THE UNIVERSITY OF MICHIGAN COLLEGE OF LITERATURE, SCIENCE, AND THE ARTS Department of Astronomy Final Report OSO SATELLITE Richard G. Teske ORA Project 05567 under contract with: NATIONAL AERONAUTICS AND SPACE ADMINISTRATION GODDARD SPACE FLIGHT CENTER CONTRACT NO, NAS5-3176 GREENBELT, MARYLAND administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR May 1970

TABLE OF CONTENTS Page OBSERVATION OF THE SOLAR SOFT X-RAY COMPONENT; STUDY OF ITS RELATION TO TRANSIENT AND SLOWLY-VARYING PHENOMENA OBSERVED AT OTHER WAVELENGTHS 1 SOLAR SOFT X-RAYS AND SOLAR ACTIVITY 13 I. Relationships Between Reported Flares and Radio Bursts, and X-Ray Bursts 13 II. Relation of Solar Soft X-Radiation to the Course of Solar Activity 35 III. Relation of 8-12 A Solar X-Ray Background to Other Indices of Solar Activity IV. Loop Prominences with Soft X-Ray Emission 83 V. Some Evidence for the Altitude of X-Ray Source Volumes in Solar Flares 101 iii

OBSERVATION OF THE SOLAR SOFT X-RAY COMPONENT; STUDY OF ITS RELATION TO TRANSIENT AND SLOWLYVARYING PHENOMENA OBSERVED AT OTHER WAVELENGTHS RICHARD G. TESKE Dept. of Astronor, and the McMath-Hulbert Observatory, The University c'ichigan, Ann Arbor, Mich., U.S.A. (Received 8 July, 1968) Abstract. Solar X-rays from 8-12 A have been observed with an ion chamber photometer and fluxes derived from the observations after an assumption concerning the spectral distribution. The time variation of the X-ray flux correlates well with the radio flux, plage index, and sunspot number. Comparisons of X-ray and optical events are given; flares seem to produce soft X-rays, but some soft X-ray bursts are apparently not associated with flares. The total energy involved in the soft X-ray bursts may be a significant amount of the total flare radiation. 1. Instrumentation The Michigan soft X-ray ion chamber photometer is located in the wheel of OSO-III. The ion chamber is filled with about one atmosphere of dry nitrogen gas and has a 5-mil thickness aluminum foil window; the efficiency of response as a function of wavelength for such an ion chamber has been published (ACTON et al., 1963). The detector responds principally to energy in the wavelength range 8-12 A, though there is also a low efficiency of response between 2-5 A. In OSO-III, the cadence of mainframe telemetry rate and wheel rotation rate results in roughly one word of solar soft X-ray data and two words of particle background data for each period of wheel rotation. Data generated by the instrument are converted to energy fluxes E(8, 12) ergs cm-2 sec-'1 for the wavelength band 8-12 A under the assumption that the radiation is distributed as in a black-body curve for 2 x 106 K (KREPLIN, 1961). If the slope of the actual flux-distribution curve departs from the assumed one (e.g., NEUPERT et al., 1967; FRITZ et al., 1967; RUGGE and WALKER, 1967), the true energy flux may be approximately recovered by applying a correction factor (Figure 1). Attempts are being made to utilize observations obtained during occultations of the sun behind earth's atmosphere near satellite dawn and twilight to determine spectral slopes for the flux distribution which may be appropriately applied to this experiment. Two automatically-selected ranges of operation are incorporated into the instrument, one with a high dynamic sensitivity (3400/1) and one with a lower dynamic sensitivity (125/1) covering a total flux range 0<E(8, 12)<0.12. The instrument has been saturated by the peak emission from great flares. An electrical calibration occurs about every 6 min, and has shown that the instrumental energy scale has been Solar Physics 6 (1969) 193-204; ~ D. Reidel Publishing Company, Dordrecht-Holland

RICHARD G. TESKE stable to about 1% since launch. The absolute calibration is probably within 6% of correct. Reliability of operation may be verified by using the sun itself as a standard source. There exists a good statistical - and physical - relationship between the 2800 MHz solar flux and the soft X-ray flux (e.g., Figures 2 and 3). Comparison of these two fluxes for April 1967 and for August 1967 separately gave very closely the same relationship. Over that 5-month interval, at least, no changes in ion-chamber response occurred. E(T)/E (T=2xlO6deg) Correction for error in adopted ^^~~1.0 \-~ ~black body temperature.8 \.6.4-.2 0 I 10 10 T (deg. K) IS' Fig. 1. If the true solar flux distribution with wavelength approximates a black-body curve for temperature T (abscissa), energy fluxes quoted in this paper must be multiplied by the value of the ordinate. 2. Observations of the Non-Flaring Sun The slow variation of soft X-ray flux with solar activity is shown in Figure 2 for the period March through August 1967. The index which is here used to characterize the quiet X-ray sun is the daily base-level, which is the lowest known flux reached during the day and is thus an index of the non-flaring sun. The X-ray base-levels are still preliminary, since complete 24-hour data have not yet been analyzed for the full six months. As may be judged from Figure 2, the soft X-ray daily base-level correlates well with other indices of the general level of solar activity: the 2800 MHz daily solar flux (p = 0.82), a McMath-Hulbert Observatory plage index (p = 0.72) and the Zurich sunspot number (p=0.78). Hence, the major component of the soft X-radiation being observed arises in the active centers (cf. UNDERWOOD and MUNEY, 1967). The 2

THE SOLAR SOFT X-RAY COMPONENT base-level enhancements of August 18/19 and of August 30/31, apparently uncorrelated with the other solar indices, coincide with East and West limb passages of a major active center on the sun (McMath plage region 8942) and apparently show it to have been strongly limb-brightened in X-radiation. In addition to its correlation with the 2800 MHz flux density, the soft X-ray background changes in a predictable way with radio flux density at other frequencies 200 150 100 50 35000 \ XAXI A AP 25000- \ AW''\ ^/ A 5000 5000 2002800 MHz 150 100.0i5E (8,12).010.005 MARC H APRIL MAY JUNE JULY AUGUST 1967 Fig. 2. Soft X-ray base level fluxes (bottom) are compared with the 2800 MHz daily flux density, with a plage index which is the sum of products of plage areas by excess intensity in the CaII K line core, and with the Ztirich daily sunspot number (top). (Figure 3). For a given variation AE(8, 12) of soft X-ray flux, the variation of radio flux density is greater at 2800 MHz, and is less at higher and lower frequencies. A linear fit seems adequate for the data so far examined. When the data for Figure 3 were assembled, no account was taken of possible center-limb effects. Figure 4 contains data for four days which are representative of a range of solar activity conditions. In the figure, each point represents a 67-second time-average of 5

RICHARD G. TESKE X-ray flux. Vertical bars represent satellite night, which is about 35 min long. Satellite day is about an hour long. Times which are given, so as to provide a time-scale, refer to the time of a satellite dawn. For May 11, May 5, and June 1, times are discontinuous across satellite night, while for April 15 the time is continuous. On May 11 quiet solar conditions prevailed: the daily 2800 MHz flux density of 104.0 was lowest since late December 1966 and no regions of significant activity were present on the visible solar hemisphere. These quiet conditions are reflected in a nearly flat, uniform X-ray flux-time curve. Arrows indicate the reported starting times for two subflares. We have associated the nearby X-ray enhancements with those flares. In contrast, on May 5 the X-ray flux-time curve was constantly disturbed, showing significant variations throughout the period of time depicted. On this date a flare-rich plage region (No. 8791) was located in the Southwestern quadrant of the disk, and 1000 MHz 2000 MHz 2800 MHz 3750 MHz 9400 MHz E (8,12) Base Level 0.010 - 0.005-'0 II I,,,II 1,I I I I I I I I I I I I I I II I I I I I I C 70 100 100 140 100 170 120 190 280 330 Flux Units Fig. 3. Radio flux densities at 2800 MHz are those reported by Algonquin Radio Observatory (Ottawa) and at other frequencies are those reported by the Research Institute of Atmospherics at Nagoya University (Toyokawa). Radio and X-ray fluxes were observed simultaneously when no flares were occurring. The Toyokawa and Ottawa data represent samples during different months. One flux unit equals 10-22 W m-2 Hz-1. the 2800 MHz daily flux density was 125.8. Arrows point to enhancements which we have associated with subflares reported during our X-ray observing hours on May 5. The flux-time curve for April 15 again refers to moderately active solar conditions, when the 2800 MHz flux stood at 123.2. It is inserted in Figure 4 to demonstrate operation of the instrument as it switches from its range of higher dynamic sensitivity to its range of lower sensitivity. The major enhancement in the diagram accompanied a reported subflare. The second enhancement was not accompanied by a reported 4

THE SOLAR SOFT X-RAY COMPONENT optical event. We have tentatively associated it with a small event in plage No. 8760 on the Northwest limb, not seen very clearly on our flare films. There occurred a weak brightening of the plage, and a surge-like, faint feature was seen projecting above the limb. Times of beginning, maximum and end of the event coincide in time with the X-ray increase. Flux E(8,12) ergs cm-' sec' 0.025 - JUNE 1, 1967 0.020-. - OOh IOm 0.015 -. II' II -' ~II II -' 11 - - ~ 16 hllh m....... 0.010'' 0.008 APRIL 15, 1967 0.004 _ -.^ 15h41m 20h 28" 0.0050r 0.00251 II II 11 11 I1105h51m 13h 49m MAY 5, 1967 0.0050 MAY 11, 1967 0.0025 -4 0I 4 III I I II I,, 0 _ 03h 40'm 11 33m Fig. 4. Representative data for four days in 1967. At times of great solar activity, such as on June 1, the solar soft X-ray flux varies almost constantly through a great range of flux values. On that date the 2800 MHz flux was 169.7. McMath plage region No. 8831, then in the Northeast quadrant of the disk, was at its peak of flare activity on June 1 and June 2. Enhancements of 5

RICHARD G. TESKE X-ray flux seen in our record for June 1 accompanied subflares and flares of importance 1. X-ray flux increases which accompanied subflares on June 1 were generally greater than those accompanying subflares on the other dates in Figure 4. Flare associated X-ray enhancements will be discussed in the next section. Our data show that during times when no reportable optical events occur on the sun, the solar soft X-ray flux may still undergo some variations. Sometimes these non-flare X-ray fluctuations can be identified with fluctuations in Ha plage intensity and with events at the limb, but often there is no clearly-associated optical Ha counterpart. At times of relatively high solar activity, as evidenced by a relatively high X-ray base-level or by a relatively high 2800 MHz flux, these non-flare X-ray fluctuations occur frequently, with an amplitude of - 10% or more of the average flux level, and on a time scale of tens of minutes. As solar activity moderates, the X-ray fluctuations also moderate as to their amplitude and frequency of occurrence. At the quietest times seen so far, the X-ray flux-time curve becomes smooth, showing only occasional fluctuations of very small amplitude. 3. Observations of Active Events The relationship of soft X-ray flux enhancements to transient solar events observed optically and at radio wavelengths is being more clearly outlined by the OSO-III data. As regards flares, the times of beginning, maximum and ending of the X-ray events, as well as the total flux enhancement, appear to be statistically related to the timing of the optical event and to its importance classification. (See, however, the discussion by UNDERWOOD (1968).) We have attempted to determine the frequency with which X-radiation accompanies flares. The study has been limited to those hours during each day for which we have already reduced data. In a sample of 57 subflares reported on a world-wide basis, 51 were definitely accompanied by X-rays. Of 47 subflares which were observed cinematographically at the McMath-Hulbert Observatory during March, April and May of 1967, 44 were definitely accompanied by soft X-rays. The three'failures' occurred when our X-ray detector was operating in its range of lower sensitivity. If the subflare sample is confined to those subflares observed cinematographically at McMath during hours when the X-ray detector was operating in its higher sensitivity range, then it appears that all the subflares were associated with X-ray emission, although at times it was very weak emission. Only six flares of importance 1 were observed cinematographically at the McMath Hulbert Observatory between March 10 and June 30 and during hours for which we already have reduced data. Each was accompanied by a significant X-ray increase. Only 23 out of 26 importance-i flares reported on a world-wide basis during hours for which we have reduced data from March 10 through May 31, had X-rays with them. None of the three'failures' was reported by more than one observing station, however, and they may be accidental reports. 6

THE SOLAR SOFT X-RAY COMPONENT Hence, we are of the opinion that probably all flares are the site of emission of soft X-rays, although at times the emission is indeed minor (Figure 4). In order for a flare-associated enhancement to be recognizable in our data as reduced so far, its duration and amplitude must be such as to yield a time-integrated flux between 8 and 12 A of a few times 1025 ergs at the sun, assuming emission into 47r steradians and no photon scattering. On the other hand, we have asked with what frequency visible flares or other active events accompany X-ray bursts. In a sample of 79 flux enhancements of all sizes recorded during March, April and May of 1967, only 57 bursts could be identified with flares reported on a world-wide basis and four with limb events such as surges leaving 18 of them, or 23% which could not be identified with reported optical phenomena. More than half of these remaining bursts occurred during European observing hours, where subflares are usually not reported as a matter of course. The average flux enhancement for the unidentified bursts was smaller than the average flux enhancement accompanying reported subflares in the same sample. Time-lapse filtroheliographic flare patrol films made at the McMath-Hulbert Observatory have been examined in connection with some of the preliminary X-ray data. The total film examined is about equivalent to 35 days of full-time day-light flare patrol. Two relatively large X-ray bursts have been found which are not clearly associated with visible flare activity; both are somewhat smaller than the enhancement usually associated with flares of importance 1. One of these may have possibly been associated with activity behind the Southwest limb (July 11). The second took place near the time of a subflare on the disk, but is in very poor time-association with the Ha event (July 29). Studies of individual flare events in connection with the X-ray data have permitted us to draw some tentative generalizations which will be investigated further as more data are reduced. Sources for the studies include the McMath-Hulbert flare patrol films, the ESSA Solar-Geophysical Data Bulletin, HAO Preliminary Reports, flare and SID lists from McMath and Manila and radio data from Toyokawa, Hiraiso, Sagamore Hill and Ottawa. X-ray enhancements associated with flares appear to be fair guides to flare importance in the sense that subflares tend to be associated with smaller enhancements than flares of importance 1, and so on, yet flares within a given importance class may give rise to a great range of X-ray flux. Generally the Ha flare and the X-ray event begin at about the same time. The Ha intensity peak will most frequently precede or coincide with the X-ray maximum. In about 20% of flares the Ha maximum intensity follows the X-ray peak. In flares of importance 1 - and 1 the Hoc and X-ray events tend to end together, while in major flares the X-rays tend to outlast the visible event for a length of time which depends upon flare importance. Following the great flares of March 22 and May 23, 1967, the soft X-ray flux remained above pre-flare levels for at least five hours. Many exceptions to these generalizations have been observed. For example, sometimes the visible flare will begin two or three min before the X-ray flux begins to increase (Figure 5). Again, 7

RICHARD G. TESKE following the importance-2 flare of May 6, 1967, the soft X-ray flux returned to its pre-flare level quite precisely and almost at once after the flare was reported to have ended. Flare events occurring close to the sun's limb sometimes demonstrate the greatest departures from these generalizations, an effect which is probably connected with the well-known decreased visibility and shorter duration of flares near the limb. Examples of X-ray bursts accompanying flares of importance 1 and 2 are shown in Figures 6 and 7, together with photometric Ha intensity curves (intensity expressed in arbitrary units) and the accompanying cm-)l bursts. In both events a slow brightening in both Ha and X-rays preceded the cm-) burst. At that time the flares abruptly brightened in Ha and X-rays. E(8,12) I Ig APRIL 4, 1967.004-,-e6 X-RAYS ~~~ ~.0031 I. 0.1 I 4 t. x SUB-FLARE *.2 -. 1.0"' -- e ". <' " ~ - ^ *''. 20h20 m 20h30m 20h40m 20h50m Fig. 5. Photometric Ha intensity curves for two subflares (lower) and the associated X-ray flux enhancements. The sun was occulted behind earth's atmosphere beginning at about 20h51m. The first subflare (S20E22) was associated with a great ejection of dark material, the second (S25E50) was not. In the July 25, 1967 event (Figure 6), the importance-1 flare lasted until after 15h00O, although the Ha light curve has not been carried along that far. The subflare which began at 14h50m on that date did not affect the X-ray record probably because our instrument was then operating in its'flare mode' of low dynamic sensitivity, in which the weak emission accompanying subflares is often not detected (cf. Figure 5, where operation was in the higher sensitivity mode). While the Ha and X-ray curves often appear qualitatively similar, as in Figures 6 and 7, the structure of the soft X-ray burst is usually very different from the structure of the accompanying cm-r burst. Typically, the maximum of the cm-) burst occurs some minutes prior to the X-ray maximum ( 80%), more rarely the two maxima coincide within a minute ( 15%) and in still rarer cases a weak cm-A maximum occurs after the soft X-ray peak. Thus, the X-ray maximum tends to associate itself most clearly with the post-burst phase of the radio emission at centimeter wavelengths. On the other hand, there is a strong tendency for the soft X-rays to begin their increase earlier than the reported beginning of the cm-) burst. 8

THE SOLAR SOFT X-RAY COMPONENT This latter effect is suggestive of physical conditions in the flare volume, to which the mechanism that gives rise to the soft X-radiation is more sensitive than is the mechanism that generates cm-A radiation. In flares, an increase of 2800 MHz flux density by 5 flux units (5 x 10-22 Wm-2 Hz- ) is accompanied by an X-ray increment AE(8, 12) of 0.005 ergscm-2sec-' (Figure 8). The relationship in the slowly-varying component (Figure 3) is such that an increase of 70 flux units at 2800 MHz accompanies an increment in X-ray flux of AE(8, 12) of 0.005 ergscm-2sec-1. WHITE (1964) noted a similar sensitivity in the 2-8 A flux. 0.100 FLUX E(8,12) JULY 25, 1967 ergs cm'-sec'' 0.050 ".*-... 0.010 4.0 / \ Imp. Ib, N28 E40 3.02.0' -..... I /,, Imp. Sb, Nil E48 1.0'I I 1170 0 M Flux Units 8800 MHz 14h20m 14h40m 15 h0m Fig. 6. The gap in the X-ray record (top) occurred during tape-recorder playback. Clouds interfered with the Ha flare observations (center). The 8800 MHz record was reported by Sagamore Hill (lower). 4. X-Ray Energy Emitted by Flares Information concerning physical conditions in flares may be obtained through knowledge of the energy invested by flares in various parts of the electromagnetic spectrum and in particle and plasma emissions. Further, the total energy emitted by flares and the time scale for its release are constraints upon possible flare mechanisms. Thus, it is of interest to obtain from our data some indication of the total energy emitted as soft X-rays. Because the spectral distribution of X-radiation is a function of time during a flare event (RUGGE and WALKER, 1967; CULHANE et al., 1963), time-integrals of flare 9

RICHARD G. TESKE energy between 8 and 12 A are uncertain, and it is difficult to estimate at this time what the uncertainty is. It may not be overly pessimistic to estimate errors by up to a factor of 3 for ordinary flares, and by up to a factor of 5 for flares of importance 2 and 3, in the sense that the measured total energies may be too high. -.0 Log E(8,12) Imp. 2b April II, 1967 - -1.5 X - RAYS -2.0 -2.5.....;.......'' Log -'. 3o Ha 0.3 ). o o.-..-' ~. 0 0 0.2-.,.. 0O~0 0 0.1 0o 00 0-1 * 00 ~ 0 C Excess Flux Units 100 80 40,.20- V - _~ 2700 MHz 20 21 hoom 21hl I 21h20m 21h30m Fig. 7. The importance-2b flare took place at S23W72. Ha intensities measured in two parts of the flare are shown on a logarithmic scale, as are the X-ray fluxes. The cm-). event is drawn schematically for two frequencies according to author's interpretation of data in ESSA Solar-Geophysical Data Bulletin. TABLE I Total Flare Energies J E(8, 12)dt Date No. of Mean Energy Range (1967) Importance Flares (ergs) (ergs) - 1- 8 3x1027 7 x 1025 - 8 x 1027 - 1 8 6 x 1028 1 x 1028 - 2 x 1029 March 22 3- 1 % 1030 (Total) May 23 2, 2, 2, 2 4 > 5 x 1030 (Total) 10

THE SOLAR SOFT X-RAY COMPONENT Peak E(8,12) Flux (ergs cm' sec-' 10~' 0 X x 0 * X 0 ". *1. Ft,Ia' a~.oPeak 2800 MHz Flux (W M'2 Hz"') o0-22 10o21 o1020 10-'9 Fig. 8. Relation of peak fluxes for associated X-ray and cm-) events. Filled circles: simple bursts with flares; crosses: complex bursts with flares; open circles: simple bursts without reported flares. All values are as measured at the earth, uncorrected for earth's orbital eccentricity. Integrated X-Ray Flux E(8,12) ergs cm-2 103. * 2800 MHz X 3750 MHz A 9400 MHz ) xXA a XA n~101- AX x A'x **x 10~ Integrated Radio Flux -Joules m-2 (Hz)-' l0-21 1o-20 1-9 Io- — 1 o6 Fig. 9. Relation of time-integrals of X-ray and cm- events. The 2800 MHz point at 5.5 x 10-20 Joules n-2 Hz-1 represents an event on July 11 which was not clearly flare-associated. The two 2800 MHz points towards lower left represent estimated upper limits based upon no report during observing hours. All values as measured at earth, uncorrected for earth's orbital eccentricity. 11

RICHARD G. TESKE We have assumed that the flare radiation is released into 47r steradians and that none of the inwardly-directed photons are scattered. Time-integrals of energy measured by us in a number of flares, and computed for the wavelength band 8-12 A are summarized in Table I. The estimates in the table should be viewed as upper limits. The energy radiated by flares at cm-wavelengths is related to the soft X-ray energy in the sense that large X-ray events tend to accompany large cm-A events and vice versa. Peak X-ray flux and peak cm-A flux density in associated events are plotted in Figure 8. We note that the X-ray peak is still high enough at a radio flux density peak of 1 or 2 flux units to be clearly discernable in our data. It appears likely that even smaller X-ray bursts than those investigated may be accompanied by unreportably small radio bursts. When time-integrals of energy at cm- and X-ray wavelengths are examined, we find again that there is a strong tendency for the greater X-ray events to accompany the greater radio events (Figure 9). Time-integrals of radio flux density in Figure 9 were obtained by multiplying tabulated mean fluxes by tabulated durations. ARNOLDY et al. (1967) have shown that a definite relation like that suggested by Figure 9 exists for the 10-50 kV X-rays. Further study of our data will be necessary before quantitative relations can be confidently established with them. References ACTON, L. W., CHUBB, T. A., KREPLIN, R. W., and MEEKINS, J. F.: 1963, J. Geophys. Res. 68, 3335. ARNOLDY, R. L., KANE, S. R., and WINCKLER, J. R.: 1967, Solar Phys. 2, 171. CULHANE, J. A., WILLMORE, A. P., POUNDS, K. A., and SANFORD, P. W.: 1963, Space Res. IV, 741. FRITZ, G., KREPLIN, R. W., MEEKINS, J. F., UNZICKER, A. E., and FRIEDMAN, H.: 1967, Astrophys. J. 148, L133. KREPLIN, R. W.: 1961, Ann. Geophys. 17, 151. NEUPERT, W. M., GATES, W. J., SWARTZ, M., and YOUNG, R.: 1967, Astrophys. J. 149, L79. RUGGE, H. R. and WALKER, A. B. C., Jr.: 1967, Preprint (Submitted to Space Res. VIII). UNDERWOOD, J. H.: 1968, Science 159, 383. UNDERWOOD, J. H. and MUNEY, W. S.: 1967, Solar Phys. 1, 129. WHITE, W. A.: 1964, in AAS-NASA Symposium on the Physics of Solar Flares (ed. by W. Hess), (NASA SP-50), p. 131. 12

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY I: Relationships between Reported Flares and Radio Bursts, and X-Ray Bursts RICHARD G. TESKE and ROGER J. THOMAS Dept. of Astronomy and the McMath-Hulbert Observatory, The University of Michigan, Ann Arbor, Mich., U.S.A. (Received 15 January, 1969) Abstract. Soft solar X-rays (8 < A < 12 A) were observed from OSO-III. An analysis of the X-ray enhancements associated with 165 solar flares revealed that there is a tendency for a weak soft X-ray enhancement to precede the cm-A burst and Ha flare. The peak soft X-ray flux follows the cm-) peak by about 4 min, on the average. Additionally, it was found that flare-rich active centers tend to produce flares which are stronger X-ray and cm-i emitters than are flares which take place in flarepoor active centers. 1. Introduction The association between optical, radio and X-ray emission from solar flares has been scrutinized recently by several observers. DONNELLY (1968) compared the timing of soft X-ray bursts with the onset and maximum of the Ha flare and with the accompanying microwave burst, concluding that there is a tendency of the soft X-ray flux to be enhanced prior to the detectable onset of other phenomena. In his study, Donnelly used preliminary data furnished by several experimenters. NEUPERT (1968) compared flux curves at 1.87 A obtained on OSO-III with radio flux-density records (2695 and 2700 MHz) for three flare events. The maximum rate of solar line emission at 1.87 A (Fexxv?) occurred 1-10 min after the peak microwave emission, while little or no change in flux from lower stages of ionization (Feix-Fexvi) took place. The data are consistent with a model in which X-ray emission is produced by thermalization of the fast electrons which give rise to the impulsive radio burst. The McMath-Hulbert Observatory has on OSO-III a soft X-ray ion chamber. In this communication we propose to describe, for 165 flare events observed by that instrument between March 9 and December 31, 1967, the statistical relations between timing of Ha, X-ray and cm-A events accompanying flare phenomena. The relationships which are found are discussed in terms of a schematic model for X-ray production prior to and during flares. 2. Observations Soft solar X-rays between 8 and 12 A were observed from the wheel of the OSO-III satellite with an ion chamber having an aluminium foil window and nitrogen gas filling (see ACTON et al., 1963). X-ray fluxes, calculated assuming a black-body energy distribution at T=2x 106 K, were observed once each rotation of the wheel (1.7-sec period) and measured in discrete steps. Up to E(8, 12)A0.0044 ergs cm-2 sec-1, the interval between steps is about 0.000035 ergs cm-2 sec-. Above E(8, 12)_ Solar Physics 8 (1969) 348-368; ~ D. Reidel Publishing Company, Dordrecht-Holland 15

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY -0.0044 ergs cm-2 sec-l, the interval between steps is about 0.00095 ergs cm-2 sec-. The instrument saturates at E(8,12)-0.12 ergs cm-2 sec'-. Graphs of X-ray flux vs. time, prepared by a computer, were made by time-averaging 6.8 sec of data for each point that was plotted, not for reasons of signal noise but purely for economy of presentation. There is no evidence that the process of time-averaging has caused loss of significant information. Because the flow of X-ray data is interrupted periodically by satellite night and by tape-recorder playback, and its quality is sometimes degraded by particle interference, we often do not observe a complete X-ray flux curve for flares. In the present investigation the data were organized around the Ha flare event as argument of entry, and cover the period March 9, 1967 through December 31, 1967. While we have examined flare-patrol films and in some instances have made intensity measures from them (Figure 1), our work here is concerned primarily with comparing our data to reports of optical and radio events. The discusssion of photometric flare light curves is left to a subsequent paper..005.125 (a).005.125 (b).004.100.004-.1.003-.075.003 A 10.1 D^>.^.^^ 0.1 -0 o^^** ^->^_-S..A.. O.l ^-.C B.. 0 0-0 A302 ~'* o~-^,.0 02.e050.001 T.025.001 -.025 APRIL II, 1967 APRIL II, 1967 -0 i I I I 1318 1338 1642 1702 Fig. 1. - (a) Importance-lb flare (N22 W24) of April 11, 1967. Abscissa: Universal Time in hours and minutes. The solid curves are X-ray fluxes. Note change of ordinate scale at 13h26m. Left-most ordinate scale is for high-sensitivity operation (prior to 13h26m) and the additional ordinate scale is for low-sensitivity operation. Units are E(8, 12) ergs cm-2 sec-1, at earth, in band-pass 8-12 A. Dots are intensity measures made in 4 parts of the region on Ha filtroheliographic flare patrol films, expressed in logioI/Jo (right ordinate scale). Io is the intensity of the Ha image at the center of the disk. Only the initial flare phases were studied photometrically. Satellite sunset occurs at 13hS53m". - (b) Non-flare brightening in the same region (N22 W24) on April 11, 1967. X-ray detector operation in high-sensitivity range only. Parts B and C of the region brightened perceptably beginning at about 16h42m. Satellite sunset at 17h03m. 14

RICHARD G. TESKE AND ROGER J. THOMAS We have attempted to limit the flare sample to single flares of importance 1 or greater. Since not all flare reports are equally reliable (DODSON and HEDEMAN, 1968), we have selected only those flares observed at 3 or more stations which were called importance 1 or greater in the Quarterly Bulletin of Solar Activity (QB). Additionally, when it appeared that the X-ray or radio data might be confused by the occurrence of two or more flares close together in time, the events were ignored. After this culling process we are left with 165 events for which we have at least partial X-ray information. Radio-burst reports have been taken from the ESSA Solar-Geophysical Data Bulletin (SGDB) and from the monthly compilations by Toyokawa and Hiraiso. As a result there is a 6-hour gap in the Universal Time distribution of our radio data, and there are fewer flares for which reports of related radio bursts were available than there are flares for which X-ray data were available. The time-accuracy of the graphs of X-ray flux vs. time is limited only by the timeaveraging mentioned earlier. In carrying out this investigation, we have read off relevant times to the nearest half-minute, since a greater precision is not warranted by the comparisons that were made. 3. Discussion of the Observations A. X-RAY ENHANCEMENT AMPLITUDE: DEPENDENCE UPON FLARE IMPORTANCE AND UPON CM-WAVE RADIO BURST AMPLITUDE Several observers have previously noted the wide variation of amplitude of X-ray enhancement associated with flares of a given importance (e.g. UNDERWOOD, 1968) which we have found in our data. In general there is a loose relationship of X-ray amplitude with flare importance and brilliance, but there also is a good deal of overlap of this relationship between the different flare categories. Our data additionally point towards a dependence of flux enhancement upon the flare-richness of regions in which flares take place. In Figure 2 we have plotted the amplitude of soft X-ray bursts for flares of different importance and brilliance categories. There is a further subdivision of the material into flares occurring in regions which were prolific flare-producers and those which occurred in relatively inactive regions. There is a clear tendency for the X-ray amplitude to increase from low values for faint importance-I flares to higher values for flares of greater brilliance and importance. Among all importance-i flares, though, that amplitude spans more than 2 orders of magnitude. Among all importance-2 flares, the span of amplitudes is greater than a factor of 10. While one of us (TESKE, 1969) previously claimed that all flares of importance > 1, and probably all subflares as well, are accompanied by at least a small soft X-ray enhancement, our flare sample here contains one event (the lowest data point in Figure 2) which apparently was not accompanied by soft X-rays. The flare was observed by three stations at N21 E80 on June 9, 1967, and was called importance 1 by two of them. It is of interest that the plage region in which it occurred (number 8843) was an old region which had had a small spot group in it for two days during 15

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY I. N on tlfig 0.0.1 0 * 0 % 0 *. 0 0 ": I E0 E_ 0.001..:. 0 0'': ~ 0~ 0.0 x0 Fig. 2. Soft X-ray flux enhancements observed in flares of importance > 1, divided according to flare took place. Another importance-n flare in the same region on June 11 was c C 0.001 =. and 9115 as'prolific' flare-producers, and plotted X-ray amplitudes for flares occurregions ring in them separately from datarolific regionser to'non-prolific' regions. (Plage regions 0.008791, 8907, 9004, 9047 and 9128 were left out of both categories, being consideredn I 2 3 I 2 3 Importance flare importance and brilliane and further divided as to whether thee flares occurred, in our estimation, in flare-rich or flare-poor active centers. flare took place. Another importa nce-n flar e i n the same region on June 11 was however accompanied by a detectable but small increase in soft X-rays (AE(8, 12)= =-0.0011). and 9115 as'prolific' flare-producers, and plotted X-ray amplitudes for flares occurregions, as independent variable. In particular, flare-richzes ou regionsuls for flare generally thosef imin flare-rich regions tend to be better soft X-ray emitters than those flares seen to occur in relatively inactive regions. A similar relationship could probably be obtained if we used, for example, the radio brightness of regions as obtained by the Stanford with strong and complex magnetic fields. 16

RICHARD G.TESKE AND ROGER J.THOMAS TABLE I Soft X-ray burst amplitude dependence upon flare-richness of plage regions, all flares of importance 1 only Percentage of flares investigated X-ray amplitude dE(8, 12) AE< 0.001 0.001 < AE< 0.01 AE > 0.01 Prolific regions 0 38 % 62% Non-prolific regions 6 % 52 % 42 % Median X-ray burst amplitude, all flares of imp 1 Prolific regions: Median AE = 0.017 ergs cm-2 sec-1 Non-prolific regions: Median AE = 0.0074 ergs cm-2 sec-1 Median X-ray burst amplitude, imp In only Prolific regions: Median AE = 0.01 ergs cm-2 sec-~ Non-prolific regions: Median AE = 0.0085 ergs cm-2 sec-1 The great range of amplitude of X-ray bursts for flares of a given importance is qualitatively like the range found by DODSON et al. (1954) for the amplitudes of 2800 MHz bursts associated with Ha flares. We have previously compared 2800 MHz burst amplitude with soft X-ray burst amplitude (TESKE, 1969) and have shown that there is a strong tendency for the two to be correlated. There is, however, a large and real dispersion. We have in Figure 3 again compared 2800 MHz burst amplitude with X-ray burst amplitude, this time using only flares from the sample under study here. Distinguishing now between events occurring in'prolific' and'non-prolific' regions, we find that flares in'non-prolific' regions are also poor emitters at 10.7 cm (Figure 3). B. TIME-RELATIONS BETWEEN IHo FLARES AND X-RAY BURSTS In examining flare reports in connection with our X-ray data, we have focussed upon start-times and times of maxima. Designation of these times by the individual flare observers is difficult at best; the reported times are often matters of the observer's judgment (DODSON and HEDEMAN, 1964). In the case of the X-ray flux curves, times of start and maximum may be designated more objectively. We have selected as starting time of the X-ray event that time when the flux curve first changes slope. Subsequent to this time there may be a monotonic flux increase or there may be fluctuations, during which the flux remains always above the pre-burst flux level. Since we are dealing with single flare events, however, the initial fluctuations, when they occur, are considered to be a part of a single-comprehensive event (see Section 4A). Ending times are often as difficult to judge for X-ray events as for Ha events. We have therefore not studied them quantitatively at this time, but merely state that it is our impression that, for flares of importance > 1, the X-rays tend to outlast the reported end-time of a flare by a length of time dependent upon flare importance and brilliance (TESKE, 1969). 17

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY 1000 * "Prolific" regions " o "Non-prolific" regions * * 0. 100*0 0 x 10- co 0 0 c Ol 0 0 0 0 0.001 0.010 0.10 AE (8,12) ergs cm2 sec&' Fig. 3. Comparison for individual flares of associated enhancements of soft X-ray flux and 2800 MHz flux. Large dots and circled dots identify events in flare-rich and flare-poor regions. 1. Starting Times At our disposal are starting times for Hoc flares listed in both the Quarterly Bulletin and Solar-Geophysical Data Bulletin. We have, however, selected from the world-wide reports the earliest starting time recorded by a cinematographic or photographic station. Often a rapid initiation of the X-ray event is reflected by a small dispersion in Ha start-times as reported by the various observatories, while a slow X-ray rise is reflected by a wide dispersion of reported Ha start-times. The latter indicates an uncertainty among stations caused by a slow rise in Ha brightness. Thus the earliest reliable optical report is preferred in making the comparison attempted here, since something must have been taking place at that time on the sun. We cannot, of course, take into account possible clock errors at the various observing stations. In Figure 4 we have diagrammed the start-time differences for flares within 60~ (heliocentric) of the disk center, that is, for t=cos0>0.5, and for flares beyond 60~ from the disk center (j=cos0<0.5). The diagram shows that: (i) there is probably a center-limb effect and (ii) the overall results depend strongly upon the sensitivity of the detector. Table II summarizes the information for starting times. In the table we also give confidence limits computed from a Fisher't' test. Where too few data points have not permitted a reasonable application of the't' test, no information is given. 18

RICHARD G. TESKE AND ROGER J. THOMAS High Range ~.> 0.5 ~' * * ****I I *6 o 0 Low Range ~ e > * 0.5 * e * ~ 0 0 I D * * * 0*- -- * - * - O Low Range 0,< 0.5 * o....... High Range a < 0.5 * -30 -20 -0 0 1 20 (Start of X-ray event) minus (Start of Ha flare), minutes Fig. 4. Comparison of reported starting times of Ha flares and associated soft X-ray enhancements. Each dot is a single event. Relative times are in the sense that negative indicates early X-rays. Arrows point to mean values. Data are divided according to instrumental sensitivity range at beginning of event and according to p, = cos 0. TABLE II Mean differences between Ha and soft X-ray starting times Sensitivity Confidence that of cos 0 > 0.5 cos 0 < 0.5 center-limb difdetector ferences are real + 1.5 min — 0.25 min Low sensitivity + 1 Low sensitivity(X-rays are late) (X-rays are early) 75 -1.3 min - 10.3 min High sensitivity 99 ~ High sensitivity (X-rays are early) (X-rays are early) 99 Confidence that sensitivity differences 95 % are real When our detector was operating in its mode of low sensitivity, where on the average an X-ray rise of AE=0.00048 ergs cm-2 sec-' is needed before it can be detected as an increase, the X-ray start tends to come after the Ha start-time. However, when the detector was operating in its high sensitivity range, in which an increase only 1/25th as great may be discerned, there appears to be a definite tendency for the X-ray rise to precede the earliest reported cinematographic or photographic Ha start-time, especially towards the solar limb. Usually the early rise of X-rays is very slow, with an abrupt increase in rate of rise at about the reported time of start of the Hao flare, or shortly thereafter. Figures 5, 6 and 7 reproduce typical flux-curves for three events. The center-limb effect found here is one in which the X-ray start-time relative to the Hao start-time becomes even earlier for flares near the limb. 19

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY.005.125.005.125.004.100.004.100.003.075.003 -.075.002.050.002.050 ~.01 025 JUNE 17, 1967 67.1.025 LY 1967 2109 2129 2149 0004 0024 0044 Fig. 5. Fig. 6. Fig. 5. A representative soft X-ray flux curve for an importance-In flare (N29 E64) on June 17, 1967. Note scale change at 21h25m. For explanation of scales, see Figure la caption. This flare has early X-rays. Satellite sunrise: 21h00m. Fig. 6. A representative soft X-ray flux curve for an importance-if flare (N29 E44) on July 25, 1967. Only two stations were reported by QB to have observed the flare, and this event is therefore not in our catalogue. Low-sensitivity operation only. X-rays appear to be late. For explanation of scales, see Figure la caption. Satellite sunrise at beginning of record..005.125.005.125..004.100.004-.100.003.075.003-.075.002.050.002-.050.001.025.001-.025 JULY 25, 1967 U I I 0 CI I I 1415 1425 1434 1449 1459 1509 Fig. 7. A representative soft X-ray flux curve for an importance-lb flare (N28 E39) on July 25, 1967. Soft X-rays show early increase at 14h18m. Low-sensitivity operation only. Sunrise and sunset at beginning and end of record. 20

RICHARD G. TESKE AND ROGER J. THOMAS Because the early appearance of X-rays has great importance in understanding the flare event as a whole, and because their early appearance may be of considerable interest in possible flare-warning devices in future space applications, we have taken care to assure ourselves of the reality of our observations. All cases where other earlier reported flares might have caused confusion have been studied and eliminated from the material used. We reserve a detailed discussion for Section 4. 2. Times of Maxima Only rarely does a flare of importance > 1 fail to carry our detector into its range of low sensitivity, so we need not distinguish sensitivity range in discussing maximum times. Figure 8 presents the time-differences found. In preparing Figure 8, reported times of maxima from the QB were used, after elimination of events with double maxima and of nearly simultaneous events overlapping in time. Table III summarizes the work. There appears to be no strong reason to believe that a center-limb effect exists. On the average, the X-ray flux maximum lags the Ha intensity maximum by about 3 min. Hoa flares occurring near the limb are apparently of shorter duration and usually are less easily visible than are flares which occur towards disk center (DODSON et al., 1956), an effect probably caused by absorption of visible radiation in the chromosphere. Because the flare-associated X-radiation is perhaps less affected by absorption p.< 0.5. 0 -.....*. 0 0 3 T e* p > 0.5:.. * *-* * - * * i I I I I.. I'1 -20 -10 0 10 20 (X-ray maximum flux) minus (Ha maximum intensity), minutes Fig. 8. Comparison of reported times of maxima of Ha flares and associated soft X-ray fluxes. Each dot is a single event. Relative times are in the sense that negative indicates early X-rays. Arrows point to mean values. TABLE III Mean differences between Ha and soft X-ray maximum times Confidence that centercos 0 0.5 cos 0 < 0.5 to-limb difference is real + 2.4 min + 3.6 min (X-rays are late) (X-rays are late) /0 21

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY processes (however, cf. WARWICK and WOOD, 1959) we seem to find that the X-ray bursts begin anomalously early towards the limb, while it is likely that the start of the flare is just hard to see. However, the relative times of maxima of Ha and X-rays for ju <0.5 and for yi>0.5 suggest that the deduced maximum time of the Ha event near the limb is not markedly affected by the process that lessens its visibility. C. TIME-RELATIONS BETWEEN CM-A AND X-RAY BURSTS 1. Starting Times Although we have selected as start-time for an X-ray event that time at which the flux-curve changes slope, the start-time of a cm-A event is judged by the radio observers as that time when the flux-density has risen through one flux unit (SGDB, Descriptive Text). We have therefore built in a bias towards recording slightly earlier start-times for our X-ray events, but it is a bias no worse nor better than some other arbitrary definition of'start-time'. We have previously noted (TESKE, 1969) that, for flares, an increment of 2800 MHz flux density by 5 flux units (5 x 10-22 W m-2 Hz-l) is accompanied by an X-ray increment AE(8, 12) of 0.005 ergs cm-2 sec-1. At the one flux unit level (see Figure 3), the soft X-ray increment is crudely 0.001 ergs cm-2 sec 1, or about one step in our low sensitivity flux records. Thus our definition of'start-time' is in reasonable accord with the cm-. definition for those data acquired in our low sensitivity mode. The data acquired in the high-sensitivity mode are about 25 times more sensitive, however. To assess relative starting times we have used the Canadian 2800 MHz or 2700 MHz reports and the 3750 MHz reports from Toyokawa. Figure 9 and Table IV High Range,u > 0.5 I' I I I I I I I I 0 Low Range L > 0.5 * ~~ 0 ~ ~ 0 I I I I I I Low Range fp < 0.5 i I I. I 0 + High Range 282 / < 0.5 -20 -10 0 10 20 (Start of X-ray event) minus (Start of cm-X event), minutes Fig. 9. Comparison of reported starting times of cm-, bursts and soft X-ray enhancements. Each dot is a single event. Relative times are in the sense that negative indicates early X-rays. Arrows point to mean values. 22

RICHARD G. TESKE AND ROGER J. THOMAS TABLE IV Mean differences between cm-2 and soft X-ray start times Sensitivity Confidence that centerof detector cos 0.5 cos0< 0.5 limb differences are real Low sensitivity 0.0 min (X-ray ate) 43 0 (X-rays are late) - 5.1 min — 22.8 mmin High sensitivity (X-rays are early) (X-rays are early) 99 % ( Confidence that sensitivity differences are real show that, when we were recording in low-sensitivity mode, the X-ray burst and cm-l burst on the average began within a minute of one another, and that there is no clear center-limb effect. Some X-ray bursts in low range in fact began quite late relative to the cm-A burst. On the other hand, data acquired in the high-sensitivity mode give a different result: without exception the X-ray increase begins many minutes earlier than the cm-)l burst. We have no confidence in the apparent center-limb difference for our higher sensitivity range, since for the sample under investigation there are only two elements in the limb sub-set. The strong tendency for a weak soft X-ray enhancement to precede cm-) bursts is undoubtedly at least partly caused by the relative sensitivities of the instruments employed in detecting the radiation. The tendency is also indicative of physical conditions in the flare volume, to which the mechanism that gives rise to the soft X-radiation is more sensitive than is the mechanism that generates cm-). radiation (see Section 4 and cf. WHITE, 1964). In those cases where an early and slow rise of soft X-rays is seen, the rate of X-ray rise usually increases near the reported starting time of the cm-) burst. 2. Times of Maxima As Figure 10 and Table V show, the X-ray maximum almost invariably follows the cm-2 maximum, by 3-6 min on the average, and there is no clear center-limb effect. Thus, the radio maximum occurs on the rising branch of the soft X-ray flux curve. On the other hand, hard X-rays reach their maximum coincidentally with the cm-) maximum (ARNOLDY et al., 1967). There are two highly discordant limb data points in Figure 10, designating cases where the X-ray maximum preceded the cm-A burst maximum by 34 min, and followed it by 31 min. Both events took place within 10 hours of one another and were both associated with events in plage region 8942 which was then on the East limb. We are inclined to consider them as being related to complex events which may have taken place partly behind the limb, since during the preceding 24 hours at least two significant X-ray events occurred in association with flares apparently going on behind the limb. Nevertheless we have not excluded the former two events from the valid data. 23

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY / < 0.5 -34 ~ * +31 ~~40 -m* *O0e O * * * *~ ~ t I I I I I I I I' I 0 / >0.5 * 0 * *@oo 00*o@ ~ S -20 -10 0 10 20 (X-roy maximum) minus (Cm-X maximum), minutes Fig. 10. Comparison of reported maxima of cm-; bursts and soft X-ray fluxes. Each dot is a single event. Relative times are in the sense that negative indicates early X-rays. Arrows point to mean values. TABLE V Mean differences between cm-, and soft X-ray maximum times Confidence that centercos 0 > 0.5 cos 0 < 0.5 cos.5cos<0.o-limb difference is real +3.1 min +5.6 min (X-rays are late) (X-rays are late) In summary we find a marked tendency for a weak enhancement of soft X-radiation to precede the reported starting times of cm-) bursts which are associated with flares. The soft X-ray maximum in flare events almost invariably follows the cm-A maximum by 3-6 min. There is a similar, marked relationship between Ha maximum intensity and soft X-ray maximum flux, with the X-ray maximum following the reported Ha maximum by about 2-4 min. On the other hand, there is a considerable dispersion in the relative starting times of Ha flare and soft X-rays, though a definite tendency towards an early enhancement of soft X-rays is also found. The dispersion in the relative X-ray/Ha starting times is not improved by making comparisons with, for example, the starting times for flares adopted in the Quarterly Bulletin. DODSON et al. (1954) have examined the relationships between Ha flare and 2800 MHz timing. They found that the two events tend to start together (on the average, the Ha flare preceded the radio event by 0.6 min), but that the radio burst reached maximum, on the average, 3.4 min before the Ha maximum intensity. We can deduce from the flare sample studied by us that the cm-) flux reaches maximum roughly 1-2 min before flare maximum, in reasonable agreement with their result' The distributions of time-differences both for the X-ray/cm-). and X-ray/Hoc maxima are significantly non-random at the 99%O confidence level. This effect is introduced by the presence in our sample of a very few events in which the X-rays reached maximum very late. 24

RICHARD G. TESKE AND ROGER J. THOMAS There appears to be no dependence of the relative timing of soft X-rays, Ha and cm-A emission upon the spectrum of the radio burst event. Reported maximum radio flux densities were plotted for each event to produce our version of a burst spectrum. The events were sorted among seven categories of spectrum shape and the relative timing investigated for each category. No clear relationship between burst spectrum and'earliness' or'lateness' of X-rays emerged. Our negative results are consistent with the conclusions of TAKAKURA (1967), who found that the spectrum characteristics of impulsive microwave bursts do not divide them into distinctly different subgroups. 4. Analysis A. STARTING TIMES: THE PREDECESSOR EVENT The foregoing discussion of our data points out that usually a weak enhancement of soft X-rays precedes the start of the cm-l event by some minutes. Apparently the soft X-rays also tend to precede the start of the Ho flare. In Figure la we illustrate the X-ray flux curve and the Ha intensity curves for the early phases of an importance-i flare on April 11, 1968. In this event, which is representative of our sample of events having early soft X-ray fluxes, the X-ray rise began about 15 min prior to the first Hoc brightening. There is no photometric evidence in Figure la that the region brightened significantly in Ha prior to the onset of the flare (which began in the position labelled D), although our curves show a minimum in Ha surface brightness just prior to 1315 UT. Photometric examination of other active centers on the disk at this time also gives no evidence of an Ha enhancement which may have been associated with the soft X-ray flux. We have therefore concluded that the X-rays which were observed prior to the flare depicted were generated by physical processes taking place in the vicinity of the plage. While cm-wave activity was recorded near the time of the X-ray event, there does not appear to be a clear relationship like that found in many other events. A weak 2700 MHz burst was reported by Pennsylvania beginning at 1310.5 UT and lasting for 5.2 min. A series of weak 2700 MHz bursts was reported to have followed, the next beginning at 1323.6, after the small X-ray peak but at the time of the strong rise in X-rays. At 2800 MHz, a burst was recorded to have started at 1335, in good timeassociation with the optical flare. In contradistinction to the lack of clear evidence for Ha brightening with the soft X-ray emission after 1315, a subtle brightening of the same plage region did accompany an X-ray event 3 hours later (Figure lb). Too meager to be reported as a flare, the Ha intensity in two posititions (B and C) increased by only about 4% of the intensity of the Hoa continuum measured at disk center. Although the soft X-ray amplitude of this event is equivalent to the amplitude of the small peak between 1320.5 and 1323.5 (Figure la) no certain Ha increase accompanied the small X-ray peak in the earlier event. It has previously been pointed out (GREGORY and KREPLIN, 1967; TESKE, 1967) that fluctuations of the solar soft X-ray flux occur frequently, often accompanying 25

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY fluctuations in the Ha brightness of parts of plages. Of the flare events which we believe were accompanied by early soft X-rays, two were available for study on our flare patrol films. One is illustrated in Figure la. The other is an importance-lb flare on July 25 (Figure 7). In both cases we were unable to find other plages on the disk which showed Ha brightening, and so have ascribed the X-ray origin to a pre-flare event. While our full sample of early X-ray enhancements might be partly contaminated by the effects of X-ray background fluctuations not associated with the pre-flare region, a study of the frequency of these fluctuations for the period of 10 March 1967 to 18 April 1967 shows that less than 20% of our flare X-ray enhancement starttimes will be in error by 1 min or more because of confusion with the background variations. Thus the majority of the early events really represent a class of phenomenon in which soft X-ray emission precedes the optical brightening. An X-ray event morphologically similar to that illustrated in Figure la is shown in Figure 11. A weak X-ray burst began at 0842 UT, reaching a low maximum at.005.125 004.100.003.075.002.050.001~ 25 MAY 15 1967 IJ I: I,, L 0815 0835 0855 0915 Fig. 11. Soft X-ray flux curve for importance-in flare (S 13 E79) on May 15, 1967. Strong particle interference occurs between 08h21m and 08h40m. Note changes of scale at 08h59m and 09h08m. Sunrise and sunset at beginning and end of record. 0844 UT. In this case, however, the earliest reported Ha start time was 0854, at the time the second soft X-ray rise began. Flare maximum was reported at 0902 UT. Because of the relatively broad wavelength band of our detector, we cannot specify with certainty the nature of the soft X-rays in the flare predecessor. We speculate, however, that the early increase of soft X-rays prior to a flare is caused by an increase in temperature in the pre-flare coronal volume, and that the X-rays are thermal in origin (KAWABATA, 1966). Solar X-ray spectra have been obtained by, e.g., FRITZ et al. (1967), NEUPERT et al. (1967) and by WALKER and RUGGE (1967). These spectra show the presence of emis26

RICHARD G. TESKE AND ROGER J. THOMAS sion lines of helium-like and hydrogen-like ions of Neix, Nex, Nax, NaxI, Mgxi and MgxIi, among others. Table VI lists the wavelengths of some of the salient lines of these ions which are observed to lie within our band-pass. No recombination continua have been identified by these observers. TABLE VI Some emission lines observed in the solar X-ray spectrum Identification wavelength (A) NeIx 11.56 13.44 13.55 Nex 9.36 9.48 9.71 10.24 12.13 Nax 11.00 11.08 NaxI 8.02 8.46 10.02 Mgxi 9.16 9.23 MgxII 8.42 Because the efficiency of our detector increases rapidly towards shorter wavelengths, down to 7.94 A, an increase in line emission from these ions and from recombination continua will be more readily detected by our higher efficiency at shorter wavelengths. However, because of the manner in which the data are reduced (a black-body energy distribution is assumed), the measured flux increase may not accurately reflect the true flux increase. Ionization equilibrium calculations have been carried out for these ions. The neglect of dielectronic recombination in the calculations is not serious, in this case, below Te 107 K, and we find that the hydrogen-like ions of Mg, Na, Ne become important constituents at temperatures between 3.5 x 106-5 x 106 K. The weak soft X-ray increase which precedes the reported cm-) event in some flares and the relationship between times of maxima are consistent with a schematic model which may be developed as follows. During the initial X-ray rise, the cm-wave flux densities may remain sensibly unchanged in a typical event (Figure 9). Following ACTON (1968) we assume that the pre-flare coronal condensation is optically thin to cm-A radiation, and so write for the 27

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY radio flux density (KUNDU, 1965) =45T12 2 -dV W m Hz- N=2 Kr5^1 N Wm"2 Hz 1, (1) A o where A0 =area of solar disk and dV= A ds, in which A = area of burst. The energy being detected in our band-pass originates as line emission (Table VI) and in the continuum. The continuum radiation is predominantly recombination emission at temperatures near 2 x 106 K (MANDEL'STAM, 1965). In our band-pass it arises principally from recombination of OvIII, Neix, Nex, Nax and several intermediate stages of Fe ionization. These free-bound emissions may be the primary contributor. FRITZ et al. (1967) observed that the continuum/line ratio was 2 in the wavelength interval 10-20 A. In the recombination continuum of an ion in the ith stage of ionization, the volume emission coefficient integrated over wavelength (ELWERT, 1954) is proportional to JFBC NZ, i+ 1NeT- 1/2 (2) The volume emission coefficient in an emission line (e.g. POTTASCH, 1964; EVANS and POUNDS, 1968) is proportional to jL oc Nz, i10- 5040WITN2 T- 1/2 (3) We may estimate Nz,i and Nz,i+l from formulae given by POTTASCH (1963), providing dielectronic recombination may be ignored and that N,<100 cm-3. If only 2 stages of ionization dominate (the ith and (i- I)th): -iz = 1{I + [Z, Ii,_, Te]}-, (4) NZ, total where 0 [Z, I, rT] = {3.8 x 310 3 ). 10- i5040i-i/Te. (5) Z 32 Noli31 Here, the symbols are as defined by Pottasch. At temperatures above Te 2 x 106 K, and for the ions of interest, 4 decreases with increasing temperature. Line emission from hydrogenic ions, and from the continua of helium-like and hydrogen-like ions, therefore increases with Te. We assume that the flare predecessor event occurs in a single homogeneous volume V, for purposes of illustration. The X-ray flux from the region, which is observed to increase, is given by [T-1/ 2N2V] EFB oC (6) NFB {1 + ( I Z, I, T]J} [T,-1/2N2V] 10- 5040w/Te EL oc. (7) {1 + [, I, T]} (7) 28

RICHARD G. TESKE AND ROGER J. THOMAS At the same time the radio flux-density remains essentially constant: S oc [Te-1/2N2V] - const. (8) Thus an initial increase in temperature in the volume is required, although expression (8) requires other, simultaneous changes. The conditions imposed by expression (8) are restrictive, but physical changes taking place prior to the flare need only follow (8) approximately. In particular we note the exponential dependence of E upon Te for the line emission, while S depends upon T,-1/2. A simple adiabatic compression of a spherical volume of radius r requires (y=5/3) Tccr-2 and Sccr-2, possibly ruling out such a compression as the flare predecessor event. Detailed theoretical calculations of the solar X-ray flux have been carried out by MANDEL'STAM (1965). His results show that in the temperature range 2 x 106 T <3 x 106 K an increase of about 10% in T, approximately doubles the soft X-ray flux between 8-12 A in both the lines and continuum. Although FRITZ et al. (1967) observed a continuum/line ratio significantly different from that predicted by Mandel' stam's calculations, it is certainly likely that the trend of flux increase with temperature predicted by him is essentially correct. The X-ray flux increment in the flare predecessor, when it occurs, is usually a fractional part of the total background soft X-ray flux, that is, the solar X-ray flux is not doubled. Because the radiation arises from a single active center, it may represent an increase in soft X-radiation from that region of a factor of 2 or more. For the April 11, 1967 event, we estimate an increase of some 200-300% from the pre-flare region alone. According to Mandel'stam's calculations, such an increase can be attributed to a temperature increase of from 200000-400000 K. This increment in T7 in the active center leads to a decrease in 2800 MHz radiation from the whole sun of a few tenths of a flux unit, if we assume no other changes take place. That the temperature increase during the flare predecessor event is not large is attested by other observations from OSO-III. NEUPERT (1968) obtained a flux curve at 1.87 A (Fexxv?) in the flare of July 25, 1967 (see our Figure 7). Our soft X-ray flux began a weak rise 9 min prior to the first enhancement at 1.87 A. NEUPERT et al. (1969) also published flux curves at 1.87 A and at 1 1.8 A (Fe xxII?) for the importance-3 flare of May 6, 1967 (see our Figure 12). Our soft X-ray flux began a weak increase 12 min prior to the first increase in the Fexxii line, which rose in intensity nearly simultaneously with the Fexxv line. Both these lines require rather more extreme excitation conditions than do the lines of Table VI. On the other hand, the Fexxii line does overlie the Oviii recombination continuum, which is expected to be strongest at Te,-2.5x 1064 x 106 K. If our interpretation of the flare predecessor radiation is correct, an initial rise in temperature of no more than 2 to 4 x 105 K occurs in the pre-flare coronal volume. Following this initial thermal stage, which may be abrupt in many events and is protracted through minutes in the class of phenomenon being described here, particle 29

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY.00.125.005 125.004.100.004-.100.003 -.075 X i.003-.075.002.050.002-050.001.025.001-.025 MAY 6,1967 MAY 6, 1967 0418 04 38 0458 0554 0614 0634 Fig. 12. Soft X-ray flux curve for importance-3n flare (S21 W35) on May 6,1967. Note change of scale at 04h33m. Record is off-scale between 05h45m and 05h00m. High-sensitivity operation only between 05h44m and 06h44m. Sunrise and sunset at beginning and end of each record. acceleration takes place. The non-thermal particles give rise to keV-range X-rays (e.g. ARNOLDY et al., 1967) and to cm-wave bursts (TAKAKURA and KAI, 1966). B. TIMES OF MAXIMA: TRAPPING OF ELECTRONS The thermalization of fast particles and their further production of X-ray emissions has been discussed by NEUPERT (1968). A further interpretation of our results offers more evidence in support of the schematic flare model put forward by him. Usually, a rapid rise in soft X-ray emission closely coincides in time to the start of the microwave burst, even in those cases where the radio-frequency radiation was preceded by a weak X-ray increase. However, the soft X-ray maximum follows the cm-) maximum by roughly 4 min. TAKAKURA and KAI (1966) attribute the impulsive cm-A burst to synchrotron radiation generated by a fast electron stream. They assume that particle acceleration ceases at the time of the burst peak, after which the burst decay is caused by loss of trapped electrons as well as by synchrotron and gyroemission losses. Electrons with small enough pitch angles may escape the trapping region on a time-scale given by the'deflection time' defined by SPITZER (1967). 1 2.4 x 1012 3 tD ----- s' sec. (9) D 87Nwcopo (p - G) In A Ne) The constant, which depends only weakly upon conditions in the coronal volume, has been calculated for Ne 2 x 1010 cm-3, Te (flare)' -108 K, Te (corona) -106 K. For the range of conditions 0.2 e< 1.0, 0101 < N < 10 cm-3, (10) 50

RICHARD G. TESKE AND ROGER J. THOMAS we calculate that 2 < tD < 240 sec. (11) We are here assuming that the electrons being lost out of the trapping region interact with denser material in the lower corona and there excite the bulk of the thermal radiation being detected by us. If the peak of the impulsive microwave burst represents the end of acceleration of electrons, which are presumed then to be lost from the trapping region by thermalizing collisions, we would expect the peak thermal soft X-radiation to occur between a few seconds and a few minutes later. The observed average time lag is consistent with ~ 1 and N- 10~1 cm-3. Because some electrons will be lost almost at once, the fast X-ray may be nearly coincident with the beginning of the impulsive radio burst. Our inability to find a relation between radio burst spectrum and differences in times of radio and X-ray maxima also seems consistent with the above scheme. TAKAKURA and KAI (1966) computed the loss rate for all electrons of energy E<1 and for different magnetic field strengths. They found that for an increase in field H from 500 to 1000 gauss, the loss-rate of electrons is not substantially increased. Thus, since the maximum in the radio flux density spectrum is directly proportional to H, we expect no relationship between lag time of maxima and the frequency of the maximum in the radio spectrum, nor is one found. Because there is also no apparent difference in lag times for bursts with and without a strong meter-wave and long decimeter-wave component, we deduce that at least the geometry and strength of the magnetic field are not critical to the production of soft X-radiation. However, they may affect the amplitude of the X-ray burst. That the locus of X-ray production is spatially associated with solar magnetic fields is apparently shown by the very important flare X-ray photographs taken by VAIANA et al. (1968). The schematic model for soft X-ray production presented here requires a close relationship between total soft X-ray energy and total cm-) energy in flares. ARNOLDY et al. (1967) have shown such a relation exists for hard X-rays. TESKE (1969) suggests that such a relation might also exist for the soft X-rays. Although our schematic model does not require that the Hoc energy in a flare be related to the X-ray energy, one of us (Thomas, unpublished) has shown that on the average the maximum Ha flux is directly related to the maximum soft X-ray flux in flares. 5. Summary Observations of soft X-rays (8 < i <2 12 A) associated with solar flares reveal that there is a strong tendency for a weak enhancement of X-radiation to precede the cm-) burst and a weaker tendency for X-rays to precede the Haoc brightening. For some flares, then, there occurs a predecessor event which in our interpretation corresponds to the heating of a volume above the active center. The temperature rise may be 51

SOLAR SOFT X-RAYS AND SOLAR ACTIVITY 2 to 4 x 105 K. Because of the requirement of expression (8) the predecessor event may not simply take the form of a compressional adiabatic heating. Following the initial gentle X-ray rise in this class of phenomenon, a more abrupt rise is seen usually in good time-association with the impulsive radio burst. We attribute the lag of X-ray maximum relative to the cm-. maximum to trapping of the high-energy electrons accelerated by the flare process itself. Thermalization of electrons escaping from the trapping region produces soft thermal X-radiation which constitutes the bulk of the radiation observed by our detector. The precipitation of electrons from the trapping region may continue for a long time following great flares. There is a relationship between X-ray burst amplitude and cm-2 burst amplitude in the sense that strong cm-A radiation is accompanied by strong soft X-radiation. Additionally, we have found that flare-rich active centers tend to produce flares which are stronger emitters of both microwave radiation and X-radiation than are flares occurring in flare-poor active centers. Acknowledgments This work has been sponsored under Contract NAS5-3176 with the National Aeronautics and Space Administration. References ACTON, L. W.: 1968, Astrophys. J. 152, 305. ACTON, L. W., CHUBB, T. A., KREPLIN, R. W., and MEEKINS, J. F.: 1963, J. Geophys. Res. 68, 3335. ARNOLDY, R. L.,K-ANE, S. R., and WINCKLER, J. R.: 1967, Solar Phys. 2, 171. DODSON, H. W. and HEDEMAN, E. R.: 1964, Planetary Space Sci. 12, 393. DODSON, H. W. and HEDEMAN, E. R.: 1968, World Data Center A Report UAG-2, August. DODSON, H. W., HEDEMAN, E. R., and COVINGTON, A. E.: 1954, Astrophys. J. 119, 541. DODSON, H. W., HEDEMAN, E. R., and MCMATH, R. R.: 1956, Astrophys. J. Suppl. 2, 241. DONNELLY, R. F.: 1968, ESSA Technical Report ERL81-SDL2, July. ELWERT, G.: 1954, Z. Naturforsch. 9a, 637. EVANS, K. and POUNDS, K. A.: 1968, Astrophys. J. 152, 319. FRITZ, G., KREPLIN, R. W., MEEKINS, J. R., UNZICKER, A. E., and FRIEDMAN, H.: 1967, Astrophys. J. 148, L133. GREGORY, B. N. and KREPLIN, R. W.: 1967, J. Geophys. Res. 72, 4815. KAWABATA, K.-A.: 1966, Report Ionosph. Space Res. Japan 20, 118. KUNDU, M. R.: 1965, Solar Radio Astronomy, Interscience, New York. MANDEL'STAM, S. L.: 1965, Space Sci. Rev. 4, 587. NEUPERT, W. M.: 1968, Astrophys. J. 153, L59. NEUPERT, W. M., GATES W., SWARTZ, M., and YOUNG, R. M.: 1967, Astrophys. J. 149, L79. NEUPERT, W. M., WHITE, W. A., GATES, W. J., SWARTZ, M., and YOUNG, R. M.: 1969, Solar Phys. 6, 183. POTTASCH, S. R.: 1963, Astrophys. J. 137, 945. POTTASCH, S. R.: 1964, Space Sci. Rev. 3, 816. SPITZER, L., Jr.: 1967, Physics of Fully Ionized Gases, Interscience, New York. TAKAKURA, T.: 1967, Solar Phys. 1, 304. TAKAKURA, T. and KAI, K.: 1966, Publ. Astron. Soc. Japan 18, 57. TESKE R. G.: 1967, Astron. J. 72, 832. TESKE, R. G.: 1969, Solar Phys. 6, 193. UNDERWOOD, J. H.: 1968, Science 159, 383. 32

RICHARD G.TESKE AND ROGER J. THOMAS VAIANA, G. S., REIDY, W. P., ZEHNPFENNIG, T., VAN SPEYBROCK, L., and GIACCONI, R.: 1968, Science 161, 564. WALKER, A. B. C., Jr. and RUGGE, H. R.: 1967, Astron. J. 72, 825. WARWICK, C. and WOOD, M.: 1959, Astrophys. J. 129, 801. WHITE, W. A.: 1964, AAS-NASA Symposium on Physics of Solar Flares (ed. by W. N. Hess), NASA SP-50, p. 131. 33

SOFT SOLAR X-RAYS AND SOLAR ACTIVITY II. Relation of Solar Soft X-Radiation to the Course of Solar Activity March 10, 1967 - May 18, 1969 Summary The performance of the Michigan OSO-III X-ray ion chamber has been evaluated elsewhere in this report. Briefly: the instrumental energy scale remained stable to within a few percent during the interval March 10, 1967, at least through April, 1968. Thus a long-term analysis of the data obtained by it and a comparison of these data with the detailed course of solar activity could be instructive and meaningful. We have, accordingly, assembled our soft X-ray data into summary form and have related it to detai led optical and radio records of solar activity obtained and compiled at the McMath-Hulbert Observatory. The comparison covers the time period March 10, 1967, through May 18, 1968. The summary X-ray records are shown graphically in Figures 1 through 9. These summaries were prepared by students from our original data, which were computer-drawn plots of X-ray flux vs. time, in the following way. The original plots were scanned by individuals who noted time-breaks that were due to (1) satellite night, (2) tape-recorder playback, (3) particle interference, or (4) other causes such as noise. For a given uninterrupted data segment the starting time, ending time, maximum flux, and minimum flux during that time segment were recorded. Information for all data segments of whatever time length were then punched on cards and used as input to a computer program which prepared the data summaries in graphical form. These summaries (Figures 1 through 9) are plotted in 27-day blocks which correspond to "Bartels days." The maximum flux and minimum flux are plotted as short horizontal lines of length equivalent to the length of the data segment, and both are joined by a vertical line placed at the middle of the time interval. Data segments shorter than 8 min were not plotted, because of the small time-scale used in the summaries. Apparent in the figures is the slow fluctuation of X-ray background associated with the general level of activity on the visible solar hemisphere. High background levels are seen during disk passages of great centers of activity in late March and early April, late May, late July and early August, mid-December, 1967, and in late January and early February, 1968. These vari35

ations and the others which are apparent in the data summaries have been associated by us with the appearance and detailed behavior of active centers on the sun. X-ray enhancements on a time scale of many hours are seen in association with great flares such as the PCA flares of March 22 and May 23, 1967, and with flares having significant gradual-rise-and.fall or prolonged post-burst enhancements at centimeter wavelengths such as on May 3, May 26, late on August 18, and on September 19, 1967. Long enduring X-ray enhancements are also seen in connection with verified major loop prominence systems such as those on June 1, and October 30 and 31, 1967. In addition there are curious, long-enduring but minor X-ray events such as those on February 22-23 and March 19, 1968 (and a few major ones like that of June 3, 1967), which have no clearly identifiable optical counterpart. Some of these may be associated with flares behind the sun's limb, as was the case early on August 18, 1967. Flare-associated spikes stand out on the summary graphs. Not all the flares during the depicted time interval are represented, since many were missed by OSO-III, and not all flare-associated spikes portray the true X-way amplitude because in numerous cases the X-ray maximum was missed. In terms of the total rate of flaring, the great regions of July-August, 1967, are well-portrayed by numerous spikes. The great t spot group of January-February, 1968, on the other hand, was not so productive of flares having significant X-ray enhancements, although that region provided the highest background flux during the 14 months analyzed here. Differences in endurance of X-ray emission following flares can be seen for the events of April 26, May 6, and May 10, and for the flare on May 3, all in 1967. The former three flares were all short-lived as X ray events, and all occurred in the same plage region, while the latter event occurred in a separate plage and was protracted over many hours. We have found that certain active centers consistently give rise to flares accompanied by short X-ray bursts while others are characterized by flares having relatively long-enduring X-radiation, These data have been discussed in extenso in a paper now being prepared. 36

FIGURE CAPTIONS Figures 1-9. Summary of solar soft x-ray flux. For each data-interval the maximum observed flux and minimum observed flux are plotted and joined by a vertical line. Data intervals shorter than 8 min have been omitted. The flux scale is logarithmic. The abscissa is divided according to "Bartels days." 37

i-d l i;, I Ii, l U~~~~~~~~~~~-~ ~~ I.7,-,.ll w lr~l-vll/7 ijli ^ 1 1 1 1 I I i i ~ i 1 28 30 1 5 10 15 20 23 RPRIL 1967:-F 1 S 10 1 I 20 25?7 MRRCH 1967 Figure 1

b T1 I'i l l~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~~~~~~~~~~T!i, ]'I i-M ii1967 ___ __ __ __ __ __1 21'~ 25301Figure 21,' ii' I,ti! I,' I 2- 25 30 1 5 10 15 20 i,1 _______________ Figure.2. Figure 2

41 1 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~I;., 1,,,,,,, 14 15 20 25 30 1 5 9 JULY 1967 0 1t i-~ i [, i,, -,.,,,,,,,,,,,,,.- [,. ~.. I J *. -17 20 25 30 1 5 10 13 JUNNE 1967 JULY 1967 Figure 3

III I11Ii i i 4 2.. —...... 6 10 15 20 25 30 2 SEPT. 1967 h-1 I_ __.10 15 20 25 30 1 5 RUGUST 1967 Figure 4

71I II 1 1t I' I1 fl _,...i-.j... - 4 30 1 5 10 15 20 25 NOVEMBER 1967 5. l I fill, plllj' I 1*.^J, L II h' I 3 5 10 15 20 25 29 OCTOBER 1967 Figure 5

11j1111~~~~~1 1114, i "b II j r~~~~~~~~~~~~~~~i~~~J~~~~ 1 1, Ihela!I 1 i I;1 11~~~~~~~~~~~~~~~~~11IAI11vih III I J Jill 11 VI' IMII 1,,1i IrII r'I,'lil I $ 0I 23 25 30 1 5 10 15 18 JANURRY 1968 __________ -F6~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~6 26 30 1 5 10 15 20 22 DECEMBER 1967 _______ Figure 6

|| "il'1,, L''Illl',,' Jl i il,,,,]I Ii'111ii.IIl[ 15 20 25 1 5 10 12 FEBRURRY 1968 MRRCH 1968'I,'-1 -1,.,%..., L9 20 25 30 1 5 10 14 JRNURRY 1968 FEBRURRY 1968 _____ Figure 7

*1-7' 1 III I-'1r4' "* % ~).1!1I1JKJWJ''I 9 10 15 20 25 30 5I *PRIL 1968 U- All P-II bll -l]Jl1 r, I' MRC 16 1 3 1 5 20 25` 30 ~ Fge MRRCH 1968 3 5 Figure 8

- - -_ _,.,,,, 30 -'- 6 10 15 20 25 30 1 MRY 1968 _ Figure 9

SOFT SOLAR X-RAYS AND SOLAR ACTIVITY III. Relationship of 8-12 A Solar X-Ray Background to Other Indices of Solar Activity March 10, 1967 - April 30, 1968 Summary For this study we have defined a daily soft X-ray background flux level as being the lowest flux recorded during ach day, and have designated it as the "daily base level" EB(8,12) ergs cm-2sec-l, as measured at the earth. This definition is dictated by the extreme sensitivity of soft X-radiation to minor activity on the sun. Since even sub-flare-associated enhancements may disturb the observed flux levels for up to a half-hour and more, and since subflaring may be nearly continuous at some times, we are obliged to seek a a background flux that may only be fleetingly observable. In practice, the base-level was measured only between the Universal hours 0600-2200, for two reasons: (1) the other indices of solar activity which were used in this study were based upon data gathered between 0500-2000 UT, generally; (2) occasionally the base level is reached at around 2400 UT, and thus a measurement obtained then would be applied to two dates without respect to changes that might take place at later times. The other indices of solar activity used in this study are listed in Table I. Except for the plage index, all were taken from the Solar-Geophysical Data Bulletins, and all corrections published in the Bulletins were applied to the data. In addition, the radio fluxes were reduced to values as observed from the earth, uncorrected for eccentricity of the earth's orbit. The plage index used is prepared at the McMath-Hulbert Observatory from daily Calcium K spectroheliograms. It is the sum of the areas of calcium plages (in millionths of a solar hemisphere) weighted individually by their intensity excess above background chromospheric intensity, Z Aj x Ij, and is an indicator of the total excess flux in the Ca II K line due to all visible plages. The daily indices for the interval March 10, 1967, through April 30, 1968, are portrayed graphically in Figures 1 through 16. In those diagrams, the data are displayed in 27-day blocks, based upon "Bartels days," and separate ordinate scales are supplied for each index as indicated on the left-hand side of the figures. The daily plage indices have been scaled down by multiplying them by 0.01. (In Figures 4 through 16, data for the first date in each diagram were erroneously deleted by the computer.) 47

TABLE I CORRELATION COEFFICIENTS FOR THE INTERVAL MARCH 10, 1967 - APRIL 30, 1968 Data Pair Correlation Data Pair Coefficient 8800 MHz / EB(8,12) 0.828 4995 MHz / EB(8,12) 0.784 2800 MHz / EB(8,12) 0.886 2695 MHz / EB(8,12) 0.863 1415 MHz / EB(8,12) 0.779 606 MHz / EB(8,12) 0.657 Rz / EB(8,12) 0.698 Z A x I / EB(8,12) 0.726 2800 MHz / Rz 0.832 E A x I / 2800 MHz 0.825 Rz /ZA x I 0.739 Correlation coefficients have been calculated from the 14 months of data and are listed in Table I. The correlations of soft X-ray background with the radio indices are in good agreement with those obtained by Wende and Van Allen for the 2-12 A flux as observed on Mariner V during a three-month interval in 1967. We have plotted the observed relationship between daily 2800 MHz flux and base-level X-ray flux in Figure 17. This figure shows that the high correlation that was obtained depends heavily upon data points measured at lower levels of solar activity. It is not clear whether the scatter of points at higher activity levels is an indication that in many instances the true baselevel of X-rays was not seen, or whether the scatter is a real one. Although the time-base of 14 months in our analysis is only about 0.1 of a solar cycle, we attempted to determine whether the data showed any indication of an evolution of the statistical relationships between indices. An indication of a secular change might, for example, suggest whether long-term changes in the total coronal emission measure occur. Accordingly, we divided our data into four partially-overlapping time intervals and calculated linear least-squares fits to pairs of the indices. The form of the regression was y (any index) = a + b EB(8,12) (1) The coefficients a and b for the four time intervals are given in Table II. We also show some of the results graphically in Figures 18 and 19. 48

TABLE II COEFFICIENTS OF LEAST-SQUARES FITS TO PAIRS OF DAILY SOLAR INDICES Time Interval Data Pair Mar. 10, 1967- June 1, 1967- Aug. 1, 1967- Nov. 1, 1967July 31, 1967 Oct. 31, 1967 Dec. 31, 1967 Apr. 30, 1968 8800 MHZ /'E(8,12) a = 263 271 286 268 b = 9057 8102 7394 11890 4995 MHz / EB(8,12) a = 130 141 156 158 b l 11080 10140 10240 13570 2800 MHz / E(8,12). a = 96.6 98.5 106 99.8 28 MZ EBb = 9822 9514 8652 11390 0 2695 MHz / EB(8,12 ) a = 94.4 96.2 99.4 98.7 Bo 8, 1' b = 9346 8561 9193 11120 1415 MHz/ EB(8,12) a= 60.8 63.9 70.8 66.6 b = 4893 4056 4126 6727 606 MHz / EB(8,12) a = 51.5 54.2 56.4 55.5 b = 2001 15300 1565 3202 Rz / EB(812) a = 33.4 45.1 55.9 54.7 b = 11240 10420 9771 10580 A x I / EB(8,12) a = 13020 6 1 0 16540 1 3400 b = 2.03 x 106 1.80 x 1.47 x 106 2.36 x 106

These regressions appear to show a slight trend which is consistent within the different pairs of indices, but which must be viewed with considerable caution. While there is fairly good agreement among the regressions for the 1967 data, the data for the time interval November 1, 1967, through April 30, 1968, suggest that, at the higher solar activity levels as indicated by, for example, the 2800 MHz flux, the background X-ray base level was below -what might have been expected on the basis of the sun's performance in 1967. We can think of three explanations for this: (1) The ion chamber lost sensitivity in early 1968. However, in-flight calibrations show no electrical deterioration, and data obtained into 1969 do not indicate that filling gas was being lost from the ion chamber. (2) The true base level was being measured in 1968 because of a decline in total flare rate, and the true base level was not always obtained in 1967 becasue of the high rate of flaring when great active centers were present on the sun. (3) The observed effect is reflecting conditions on the sun associated with the dramatic decline in rate of flaring after midFebruary, 1968. There are also statistical reasons for caution about inferring evolutionary trends from the regressions. The linear form that was chosen is only an approximation to the true relationships (see, e.g., Figure 17). Further, the scatter of data points at h4igher levels of activity introduces an error into each coefficient a and b. Auto-correlation curves have been calculated for EB(81,2), RZ, Z Aj x Ij, and the 2800 MHz daily flux, see Figures 20 through 23. While the latter three indices show a negative minimum in their auto-correlations at lags near 14 days, the EB(8,12) auto-correlation does not dip to negative values but shows a small positive minimum near a lag of 16 days. One possible interpretation of this is that the X-radiation arises from levels high enough in the corona to be easily detected even when the underlying plage is approximately one day's rotation beyond the sun's limb. As regards the relative amplitudes of the maximum near 25 days and the minimum near 42 days, the X-ray curve behaves most nearly like that for the plage index, emphasizing the close association of soft X-ray emission with Ca II K structure which is revealed by direct X-ray photography. Again, however, the positions of the maxima and minima in these two curves suggest a high atmospheric level for the X-ray source volume. Cross-correlation analyses have been carried out for lags between minus and plus two solar rotations. These are displayed in Figures 24 through 29. The sense of the lags is given in the figure captions. These cross-correlations reproduce for the solar X-radiation the familiar influence of sunspots upon the extent and lifetime of their associated plage regions, in the sense that the more extensive spot groups, which strongly contribute to Rz, are usually associated with larger, brighter, and longer-lasting plages (see Figures 25, 27, and 29), and are also thus associated with the most prominent X-ray source volumes. It is tempting to infer, 50

from the amplitudes of the lagged peaks at + 28 days in the cross-correlation of X-rays with the 2800 MHz flux (Figure 24), that the development curve of X-ray emission from evolving active centers tends to have a phase lag with respect to the development curve for the 2800 MHz radiation. However this effect, if real, is small. 51

FIGURE CAPTIONS Figures 1-16. These diagrams graphically portray the daily variations of solar activity indices during the interval March 10, 1967 to April 30, 1968. The abscissa is divided according to "Bartels days." Separate ordinate scales are given for each quantity; all scales are linear. Generally the scales are such that the X-ray data appear as the lowest curve and the 606 MHz daily flux appears as the highest curve. (Beginning with Figure 4) data for the first date of each graph was accidentally omitted.) Radio fluxes are given in units of 10-22 watts m-2 Hz-1. Figure 17. The daily base level of soft X-ray flux is here plotted against the daily 2800 MHz flux level. The radio flux density is given in units of 10-22 watts m-2 Hz-1. Figure 18. Least-squares linear fits to segments of 14 months of data are shown. Here, the X-ray base level is compared with the daily McMath-Hulbert plage index. Figure 19. Least-squares linear fits to segments of the 14 months of data are shown. Here, the X-ray base level is compared with the daily 2800 MHz flux. Figures 20-23. Auto-correlation curves for daily solar activity indices and for the soft X-ray base level. Figure 24. Cross-correlation of soft X-ray base level with daily 2800 MHz flux. Negative lags correspond to placing X-ray data upon later dates of the 2800 MHz data set. Figure 25. Cross-correlation of soft X-ray base level with Rz. Negative lags corresponding to placing X-ray data upon later dates of the Rz data set. Figure 26. Cross-correlation of soft X-ray base level with plage index. Negative lags correspond to placing X-ray data upon later dates of the Z A x I data set. Figure 27. Cross-correlation of 2800 MHz daily flux with Rz. Negative lags correspond to placing 2800 MHz data upon later dates of the Rz data sets. Figure 28. Cross-correlation of the 2800 MHz daily flux with plage index. Negative lags correspond to placing 2800 MHz data upon later dates of the Z A x I data set. This curve is normalized at zero lag. The maximum occurs at a lag K = -1 day. 52

FIGURE CAPTIONS (Concluded) Figure 29. Cross-correlation of plage index with Rz. Negative lags correspond to placing Z A x I data upon later dates of the Rz data set. 53

X-RFA TLUX X10.00.05.10.15.20 2800 MHZ FLUX 00.00 10.00 160.00 210 00 260.00 310.00::::.;,;,,,,.,,,,:,,:, t:' SUNSPOT NUMBER A.00 50.00 100.00 150.00 200.00 250.00 PLfGE INDEX X.01 X.00 100.00 200.00 0.00 40.00 500.00 600.00 8800 MZ FLUX.00 50.00 100.00 SO.00 200.00 250.00 300.00 350.00 400.00 4995 MHZ FLUX X.00 500 10000 0. 00 200.00 250.00 300.00 350.00 2695 MHZ FLUX I.00 50.00 100.00 150.00 200.00 250.00 1415 8M1i FLUX X.00 50.00 100.00 150.00 200.00 606 MHZ FLUX *.00 50.00 100.00 150.00 (D D-O Co.~ \ 1/i ~ * ]'\ \\{\::]:: 1' / ) I t -,,J~~~~~

X-RAY FLUX XIU m.00.0O.10.1.20 2800 MIZ FLUX o dO.00 110.00 10.00 210.00 200.00 310.00 SUNSPOT 1NUBER8.00 50.00 100.00 10.o0 200.00 250.00 8600?t0OFLUX PLAGE INOEX X.01 X 100.00 E00.00 100.00 W0.00 500.00 0 00.00 8800 HZ FLUX.00 50.00 100.00 150.00 200.00 250.00.00.00 350.00 00.00 0995 MZ FLUX.00 50.00 100. 00 200.00 25.00. 00.00 0000 50.00 2695 MHZ FLUX V.00 50.00 100.00 ISO. 00 200.00 -,50.00 1415 MHZ FLUX x.00 5.00 100.00 100. 00 00.00 606 MHZ FLUX *.00 50.00 100. 150.00 F _ s + _ e e _+ @ -4,-4- / -3','1-,1.._

95 X-Y FLUX XIO 0.oo..,o............. 28009 Z FLUX 0 60.00 10.00 1,60.0 210.00 20.00 310.00 SUNSPOT NUMBER.00 50.00'1p0.00 ISO.00 200.00 250., 00 PLAGE INOEX X.01 X.00 100.00. 20.00 00 00.00 o o.oo. 00.00 00.o00 880 HZ FLUX,.00 50.00 100.00 t150. 200.00 20.00 300.00 350.00 O 0000 4995 MHZ FLUX X.00 0.00 I1O 0 150.00 20000.00 o.0 300.00, 350.00 2695 MHZ FLUX.00 0S.00 100.00 150.00 200.00 2 SO.OO o 1415 MHZ FLUX X.00 5.00 100.00 150.00 200.00 606 M12H FLUX.W SO.OO 100.00 150.00.00 50.00 100.00, 0 I t i,. I. 1 4 i i i 4 1) / y / ), 9 Ln,~ ~i,.,l u U* 0:

L2 X-RMY FLUX XI0 m.00.05.10.13.20,,, 2800 HHi FLUX O 80.00 11. 10.006 210.00 260.00 310.00.00 50.00 100.00 150.00 00.00 250.00,,,,, PLRGE INOEX X.01 X.00 10p.00o 200.00 3 O.00 40o.00 500.00 600., 00 8800 MHZ FLUX.00 50000 1 0.00 200.00 250.00 300.00 00 0.00 4995 MHZ FLUX X.0 0 S000 100.00 150.00 200.00 250.00 300.00 350.00 2695 MHZ FLUX I.db 50.00 100.00 1 50.00 00.00. 00 1415 MHZ FLUX X.00 soo,100.00 10.00 200.00 606 MHZ FLUX *.00 50.00 100.00 1t0.00 F ______. I 4__ +__-. +..-. —.-. __- -- - -.^ --- --- ---- - 4 ~ 4 / / CLI) I F' *~/ // f — +_/ 4 i / /t L..

8 U- 8 U. E 8 8 8 8 8 8 8 8 8 8. -.. JUNE 1967 |U T I6. 8 8 S8 8 I8 8 8. s Si 8 Si 8 Figure 5

6^ X-flY FLUX X10.00.05.10...20 2800 MHZ FLUX 0 60.00 I1. 00 160.00 210.00 260.00 310.00 SUNSPOT NUMBER a.00 50.00 100.00 150.00 000.00 250.00 PLAGE INDEX X.01 X.00 100.00 200.00 300.00 400.00 50.00 600.00 8800 MHZ FLUX 00 50.00 100.00 150.00 200.00 250.00 300.00 350.00..00.O 4995 MHZ FLUX x.00 50.00 100.00 150.00 200.00 250.00 300.00 3s.O. 00 2695 MHZ FLUX.00 50.00 100.00 150.00 200.00 250. 00 1415 MHZ FLUX x.00 50.00 100.00 L5000 000.00 608 MHZ FLUX * 00 50.00 100.00 150.00 c-n'in i l o- - r_ \ / \\\ Fl'j~~~~~~~~~~~~~~~~~~~~A (D CD (r\< o i ^ u Ln wo C) CC) Ln'- f A / T- [ *-1 77 -~~

8 O x e x ~ 8 8.~ L.. I u. u. 8 8, ^ 8 8 8 8 8 8 8 8 I 8. _______________________________ __________.. 10 15 20 25 30 1 5 RUGUST 1967 Figure 7

8 B 0 X x X 8.8 8 8 I C 6 10 15 20 25 30 | 8Si 8 F 8 8 F Si Si 8 8 S.. ______________________________________________ Figure 8

39 X-RAY FLUX XIO B.00.05.20.5.20 2800 MhZ FLUX 0 60.00 10.00 160.00 210.00 260.00 310.00 SUNSPOT NU6MBE A.00 50.00 100.00 150.00 200.00 250.00 ~~~~~~~~~~~~~~~~~~ I i i i i i i 4 i L — PLAGE INDEX X.01 X.00 100.00. 00.00 000.00.00 0.00 600.00 0000 800 MOZ FLUX 0.00 50.00 100.00 150.00 200.00 2S0.00 300.00 350.00 000.00 099S MHZ FLUX X.00 50.00 100.00 150.00 200.00 250.0 300.00 350.00 2695 MHZ FLUX N.00 50.00 100.00 ISO.00 2O0.00 2SO. 00 I.... — ~ ~ 0 -4-0 LOQ 50.00 200.00 250.00 1415 MHZ FLUX ---- --- ^ *.~00 50.00 100.00 150.00 2 10.00 606 MHZ FLUX f 00 50.00 100.00 150.00 CO:JJ i ) w~~~~ RI 1 I Lu Ha (D \10 oi \ \, n CD3 CD~ m~~~~~~ CT) R. C)

G9 X-RRA FLUX X10 L1.00.05.10.15.20 2800 MHZ FLUX O 60.00 110.00 160.00 210.00 260.00 310.00 SUNSPOT NUMBER &.00 50.00 100.00 150.00 200.00 250.00 PLAGE INOEX X.01 X.00 100.00 200.00 300.00 400.00 500.00 600.00 8800 6HZ FLUX.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 400.00'1995 MHZ FLUX.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 2695 M4Z FLUX.00 50.00 100.0000 15.00 200.00 250.00 1415 HHZ FLUX X.00 50.00 100.00 150.00 200.00 606 MHZ FLUX *.00 50.00 100.00 150.00: c LO CD I _ _ _ __ ___ C:1

19 X- YT FLUX XIX0 9.00.0os.10.15.20 0 2800 HZ FLUX 0 0.00oo 110.00 160.00 210.00 280.00 310.00 SUNSPOT NUMBER r.00 50.00 100.00 150.00 200.00 250.00 PLAGE INDEX X.01 X.00 100.00 200.00 300.00 0OO.00 500.00 s 00 i 8600 MHZ FLUX.00 50.00 100.00 100.00 200.00 250.00 300.00 000.00 60.00 01995 MHZ FLUX X.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 4 4~ ~ ~ ~ 2695 MHZ FLUX.00 50.00 100.00 150.00 200.00 250.00 1615 MHZ FLUX X.00 50.00 100. 500 5OO |00.00 606 MHZ FLUX *.00 50.00 00.00 so 0.00 a) Ln C-) rn~~~~~~~~~~ w H m 0x / CDC 0 m CT.).n: F;I

~9 X-RRT FLUX XIO 0.00.05.10. 1.20 2800 HZ FLUX 0 SO.00 LO~0. 160.00 210.00 2.000 310.00 SUNSPOT NUI4BER.00 50.00 00.00 150.00 200.00 0., 00. PLCGE I0lEX X.01 X.00 100.00 200.00 300.00 WO.00 500.00 0.00. 8800o HZ FLUX.00 50.00 100.00 150.00 200.00 250.00 300.00 3i50.00 00.00 49S5 2Z FLUX X.00 i0 o00 1.00 200.00 250.00 300.00 350.00 2695 4tZ FLUX X.00 50.00 100.00 tSO.00 200.00 250.00 1415 0MZ FLUX.00 50.00 o00.00 150.00.:00 606 JZ FLUX *.00 50.00 100.00 150.00 ru~~/ C) Ln, DD4 c' co C l ES/ I

99 X-RAY FLUX XIO [.00.05.10.15.20 2800 NIH FLUX O.60 110.0 O00 16 20.00 2 00 260.00 310.00 SUNSPOT NUMBER &.00 50. 10000 00 Iso.0o 200.00 z50.00 PLAGE INDEX X.01 X.00 100.00 200.00 300.00 400.00 500.00 600.00 -4 i i I 4I -4 4 4 8800 tHZ FLUX.00 50.00 100.00 50. 00 200.00 250.00 300.00 350.00 4o00.00:: T -—!- -— ~ —— + —-— +-. — + I —— + -— +- 4 -I+-a -— +I - 4 - 4995 m1Z FLU/ X.00 50.00 L00.00 150.00 200.00 2zO.0 300.00 350.00:.;: - -- ----- - I- 4 1 —-- I - 2695 AHZ FLUX B.00 50.00 100.00 150.00 200.00 250.00: ~_ - ~:+ 0 + - — + —-_ I — + —- -+ ---..+ - -.4 1415 MHZ FLUX X.00 50.00 100.00 150.00 200.00 - 4-4- -— + —-. - 4 -....: - ------------ -- — + —-- — +-.. --- ------- ------------- 606 MHZ FLUX *.00 50.00 100.00 150.00 ~ -I --- 1 —- ------------— +- --------------- ---------------------— +- -.-+ —-—. — H -' V V\~~~, t 3D C: _ s~ ~'^ ~ -- i/_/ i

L9 X-RAY F.UX X10 2.00 05.10.Is.20 2800 MHZ F LUX 1 60.00 110.00 160.00 210.00 260.00 310.00 SUNSPOT NUMBER A.00 50.00 100.00 150.00 200.00 250.00 PLRGE INDEX X.01 X.00 100. 20000 300.00 4 00.00 500.o 600.00 8800 MHZ FLUX ~.00 50.00 i00.00 150.00 200.00 250.00 300.00 350.00 00.00 4995 MHZ FLUX x.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 2695 MHZ FLUX X.00 000 50 0 150.00 200.00 250.00 1415 MHZ FLUX X.00 50.00 100.00 150.00 200.00 606 MHZ FLUX *.00 50.00 100.00 150.00 mcn cJ' In LO II -'~ i' * t

99 X-RRY FLUX X10 I.00. 0.10. 20 2800 It F.UX 0 60.00 110.00 160.00 210.00 250.00 310.00 SUNSPOT NU1HBER.00 50.00 100.00 150.00 200.00 250.00 PLRGE INOEX X.01 X.00 100.00 200.00 300.00 000 SOO.00 800.00 8800 HHZ FLUX 0.0 0. 0.00 150.00 O. ISO.00 200.00 2S0.00 300.00 350.00 W00.00 4995 MZ FLUX X.00 S0.00 100.00 150.00 200.00 200.00 300.00 350.00 2695 MZ FLUX.00 50.0 100.00 S O.00 200.00 250.00 1415 1Z FLUX z.o _.oo._.00 50.00 100.00 150.00 00.00 606 MHZ FLUX *.00 50.00 O10.00 1I0.00 i3. L, U, Ca) I II " (/ <:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

69 X-RAY FLUX X10.00.05.10.15.20 2800 MHZ FLUX ~ 60.00 110.00 O160.00 210.00 260.00 310.00 SUNSPOT NUM1BER.00 50.00 100.00 150.00 200.00 250.00 PLRGE INDEX X.01 X.00 100.00 200.00 300.00 400.00 500.00 600.00 ~ ~ -- — ~~~~~~~~~i i i & - 4 4 8800 MHZ FLUX.00 50.00 100.00 150.00 000.00 250.00 300.00 350.00 400.00 4995 MHZ FLUX X.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 2695 MHZ FLUX I.00 50.00 100.00 150.00 200.00 250.00 1415 MHZ FLUX X.00 50.00 100.00 150.00 200.00 606 MHZ FLUX.00 00.00 100.00 150.00 s~~ ~~~~ ~~~~~~~~~~~~~~~~~~~~~~~ 4l i S i -4 LO 09 C D Ln C~~~~~~~~~~~~~~~~~~fl~~~~~~~~~~~~~~~f 3D C) A lh 1 A t i y \\\ sUV { {1 I 0 VI

0.020 0.018 0.016. 0.014 C) 2 0.012 C) 0.010 y 0.008a) **.I.- * * ~.@@. *. *... * I a, 0.006 t'' | 0.004,' ~:.':I'.%~;.. I::-:.... l..~'..:...;.. * *..: *.-.:.! s o re: -' 0.002-:.'ri' 0.000 80 100 150 200 250 2800 MHz DAILY FLUX Figure 17 70

.015 co0 8/1/67 i /// u ~wy^^~- 1/1/31/67/67 Fj < // -4/30/68.0010/ I 3/10/67 / - 7/31/67 6' 6/1/67 ^/.-10/31/67 0 10,000 20,000 30,000 40,000 2 A x Iexcess Figure 18

^i.010. 00 -J/ 3/10/11/1/67 -7/31/67./ -4/30/68 U) m.005 8/6/67 -12131/67 3/10/67 - 7/31/67 6/1/67 // -10/31/67 0 100 200 2800 MHz FLUX Figure 19

1.0 0.8 0 z 0.6 Q: o 0.4 o 0 0.2 o. < 0.0 -0.2 -0.4 I I I I I 0 10 20 30 40 50 60 K = LAG IN DAYS Figure 20

1.0 z 0.8 o 0.6 0 0.4 a: 0.2 z 0.0 0 c -0.2 Z cn -0.4 0 10 20 30 40 50 60 K LAG IN DAYS Figure 21

1.0 0.8 o 0.6 D. 0.2 8 o4\ 0: -0.2 --— 3 -0.4 -0.6, 0 10 20 30 40 50 60 K = LAG IN DAYS Figure 22

1.0 w z 0.8 0.6 0 0 0.4 X 0.2 - j ON 0.NQ 0 -0.4 0 10 20 30 40 50 60 K LAG IN DAYS Figure 23

1.0 x 0.8 uI N 0.6 0 0.4 00 (O 0.2 u) <: 00 x z iw-0.4:: o -0.6 03 U) U):: -0.8 -1.0 I I I I I I I -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 K=LAG IN DAYS Figure 24

1.0 0.8 0.6 CJ c: o 0.4 c) 0 I 0.2-:D z -0.2I0 a. 03 -0.4e -0.6 1-0.8 - I. 0 IIII IIIII -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 K= LAG IN DAYS Figure 25

1.0 0.8 z 0, 0.6 w Q: 0.4 C,) n u0.20 x O0.0z_ w -0.2a. -J x. 0.4-0.6-0.8 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 K = LAG IN DAYS Figure 26

1.0 z 0 0.8 LJ O: 0.6 0 o3 0.4 0 cn Z-0.o cn -A.4 - x o3 -0.6 0 N'-60-50 -40 -30 -20 -10 0 10 20 30 40 50 60 K= LAG IN DAYS Figure 27

I.c z 0 p0.8.I r: 0.6 011 C/) (.) 0.2 - -0.00.I. -0.4 - 0_ x -0.4 - o-0.8 - 0 600 10 20 30 40 50 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 K= LAG IN DAYS Figure 28

1.0 z 0 P 0.8t-J cs 0.6 0 0 tI Cn 0.40 - 0 r -0.2z 0.0 0 a. C) -0.2 - z -0.4 a- -0.8 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 K=LAG IN DAYS Figure 29

SOFT SOLAR X-RAYS AND SOLAR ACTIVITY IV. Loop Prominences with Soft X-Ray Emission I. Introduction Great systems of loop prominences are closely associated with major flares. They were first identified as flare-associated structures by Dodson (1961). Bruzek (1964), in a classical observational study, investigated the flare/loopprominence relationship and described the loops as a "prolonged flare process which, slowly decaying, propagates into the solar corona." The major loop systems which Bruzek described appear within sporadic coronal condensations following the onset of great flares. The flares are themselves found to be the sites of emission of Type IV radiation and are often productive of subsequent PCA effect. Jeffries and Orral (1965a, 1965b) suggested that the loop prominences have their origin in the thermalization of fast particles which are guided along magnetic field lines. In their view and in Kleczek's (1964), at least a significant fraction of the loop prominence material is delivered to the system in the form of protons of keV and higher energies. The delivery of these particles onto the loops, if it occurs, is unobservable at visual wavelengths, but the process may be expected to be observable in the X-ray region of the solar spectrum. Olson and Lykoudis (1967) have considered an alternative model for production of loop systems based upon a condensation process. In their hypothesis, plasma confined to the coronal condensation is compressed and subsequently loses energy by radiation and by conduction, becoming visible in Ha as loops. Because no material is added to the system it is unlikely that this mechanism can give rise to a major loop system, since the observations require that the loops contain more mass than is available from the coronal condensation (Kleczek, 1964). Olson and Lykoudis concluded that loops produced by their mechanism would comprise small systems h i 30,000 km, and that these would be insignificant emitters of X-radiation. We have obtained excellent records of the solar soft X-ray flux between 8 and 12 A covering the period March 9, 1967, to June 28, 1968 (Teske, 1969). These records were examined for soft X-radiation that may have been associated with major systems of loop prominences, with the aim of understanding something further about the processes which give rise to those structures. 83

II. The Data Major systems of loops are rare phenomena: Bruzek (1964) assembled twenty-nine cases of major systems at the limb during a seven-year period November, 1956-September, 1963, without commenting upon the completeness of his list. A list was made by us of the twenty-six loop systems reported at the solar limb by Sacramento Peak (in Geophysics and Space Data Bulletin: GSDB) during the interval March 10, 1967-December 31, 1967. A twenty-fifth case (May 21, 1968) was added to the list because of its reported magnitude: the loop tops extended to 0.08 solar radii from the limb. Seven additional cases of flares accompanied by loops were culled from the Quarterly Bulletin on Solar Activity (QB) for the same time period. A reasonably prominent soft X-ray enhancement was found to accompany only ten of the thirty-three events in the catalog (Table I), although several others may have been associated with very weak soft X-ray enhancements of amplitude AE(8,12) < 0.005 ergs cm-2sec-1 at the earth. In our opinion, weak X-ray enhancements are easily confused by subflaring activity and with fluctuations in the background flux, and so these have not been studied. Our flare patrol films are not especially suited to the examination of loop prominence images. Where possible we have examined them to confirm-the identifications. Our impression is that the majority of the "loops" reported in GSDB are indeed loop-like configurations but are not the major systems described by Bruzek. Thus the low fractional association of reported "loops" with soft X-radiation may have no meaning with respect to the great systems. We reserve comment as to whether the majority of the reported "loops" may be identified with phenomena generated by the Olson-Lykoudis mechanism (1967). Optical (Ha) data available to us permit the reasonable identification of four of the ten "yes" cases of Table I as major loop systems (Table II). These four events are discussed below. In Section III we will consider those features of our observations which are relevant to the compression model and to the injection model. 84

TABLE I REPORTED Ha LOOPS Date Time, UT Source Position Soft X-Ray Enhancement 3/18/67 1510-1550 GSDB N10 W90 No 3/20/67 1322-1542 GSDB N26 E90 Yes 3/21/67 2022-2034 GSDB N16 E90 Possible 3/22/67 0022 QB N24 E68 Yes 3/27/67 0441 QB N24 W14 Weak 4/ 2/67 0407 QB N24 W86 No 4/ 2/67 2054-2159 GSDB N18 W90 No 4/14/67 1532-2035 GSDB S21 W90 Yes 5/18/67 1927-1958 GSDB N26 E90 Moderate 5/18/67 1934 QB N25 E80 Moderate 6/ 1/67 1214E-2040U GSDB N24 W90 Yes* 8/ 4/67 1526-1603D GSDB N30 W90 Moderate, short-lived 8/ 6/67 1400-1442 GSDB S24 E90 No 8/18/67 2044-2129 GSDB N24 E90 Yes 8/23/67 0515 QB N23 E36 No 9/24/67 1444-1652 GSDB N13 E90 Possible? 9/24/67 2152-2242 GSDB N20 E90 No 9/26/67 1846U-2014 GSDB S23 W90 Weak 10/ 3/67 0943 QB S20 W83 No 10/19/67 1920-2014 GSDB N14 E90 Possible 10/27/67 1434-1645 GSDB S15 E90 No 10/27/67 1758-1830 GSDB S17 E90 No 10/27/67 1946-2036 GSDB S13 E90 No 10/27/67 2232-2310 GSDB S16 E90 No 1.0/29/67 2128U-2235D GSDB N10 W90 Yes 10/29/67 2347 QB N10 W90 Yes 10/30/67 1654E-2400D GSDB N10 W90 Yes* 10/31/67 1912-2352D GSDB N12 W90 Yes* 11/ 3/67 1415-1442 GSDB N20 W90 Possible 11/ 4/67 1635-1811 GSDB N17 W90 No 11/10/67 1338-2342 GSDB S25 W90 Possible? 12/ 2/67 0810 QB N21 W90 Possible 12/ 2/67 1426-1702 GSDB S28 W90 Possible 12/ 4/67 1302 QB S19 W90 Yes 5/21/68 2050-2440D GSDB N16 E90 Yes*. Optically identified as major loop system. 85

TABLE II LOOP PROMINENCES FOUND TO HAVE A SIGNIFICANT SOFT X-RAY ENHANCEMENT X-Rays X-'Rays - X-Ray Peak Flux MH Loops Date Begin End cm2 secu First Visible T Comments (UT) (UT) AE(8,12)ergs cm-2sec*1 First Visible (UT) 6/ 1/67 0700 1500 0.022 Unknown 10/30/67 <1740 >2400 0.015 1850 10/31/67 1550 >2500 0.013 1912 5/21/68 <2020 >2400 >0.016 2041 Poor Ha data A. OCTOBER 30 AND 31, 1967 Data relating to the time-period October 30 (12h UT) to October 31 (24h UT) are presented in Figure 1. We have associated the two prolonged X-ray enhancements on each day with large systems of loop prominences seen at the solar west limb. Ha filtroheliograms of these systems are shown in Figure 2. These were very kindly made available by Dr. Hyder at Sacramento Peak (Figure 2a and 2b, October 31) and by Mr. Miller at Lockheed California Company (Figure 2c, October 30). Iyder (1969) estimates that the top of the system on October 30 had reached an apparent altitude h - 1.2 x 10 km by 23hl5m UT, while on October 31 the loop tops appear to have reached a projected altitude h - 4 x 104 km by 19h24m UT. From a consideration of the probable location of the flares which may have initiated these phenomena (Teske 1970) we have deduced actual heights of the loop tops h; 105 km. These systems thus rank among the largest studied by Bruzek (1964). Neither of these systems was associated with Type IV radiation, an observation to be compared with Bruzek's (1964), in which only four out of twentyseven cases of loops at the limb were accompanied by recognized Type IV (two cases occurred prior to Type IV's recognition). However, the phenomena seen in October, 1967, appear to be ranked with major loop systems on basis of morphology and magnitude. There is no evidence for an X-ray burst associated with an initiatory flare for either loop event (Teske, 1970), and so the soft X-rays being observed were very likely entirely associated with the loop systems themselves. The first loops on October 50 were seen on McMath-Hulbert films at 18 50, about 35 min before soft X-ray maximum. On October 31, Sacramento Peak saw visible loops beginning at about 19h12m, perhaps 15 min after the (approximate) 86

maximum X-ray flux. The time of appearance of a loop system is probably an observational artifact, since the systems grow in height from low mounds (Bruzek, 1964), and these times correspond in our present observations to the times of first recognition of the prominences as a system of loops. The differences in time of loop recognition and time of X-ray maximum may be reconciled if we assume (i) that on October 31 the system had to rise to a considerable altitude (h - 60,000 km, Teske, 1970) before becoming visible from the earth and (ii) the X-ray source is located at altitudes above the limb and thus above the Ha source. According to Bruzek, loop systems require > 1h to grow to 60,000 km altitude. Thus, on October 31, recognizable but smaller Ha loops may have been invisibly present'below" the solar limb, prior to X-ray maximum. The active center with which these two loop systems were associated (McMath region No. 9034) became exceedingly active active on October 29, while it was on the west limb. An importance 2b flare which began at 2347 UT on the 29th may have been associated with a system of loops, according to the Quarterly Bulletin on Solar Activityo Our satellite data show the flare X-ray burst and a lingering strong enhancement of soft X-ray flux following the flare; the flux declined slowly over an interval of 6-8 hr, from an initial post-flare level comparable to the levels observed with the other two October loop systems. The flare itself was accompanied by strong cm-A radiation and by a strong postburst phase lasting more than 3 hro No Type IV was reported. Thus this active center produced, during a 48-hr interval, three systems of loop-type prominences, none of them associated with reportable Type IV radiation. We have not studied the loops of October 29. We have no optical data concerning them other than the comment in the Quarterly Bulletin. It is noteworthy that the loops were reported by Sacramento Peak as beginning before the importance 2b flare (Table I). Thus the loops may have had their origin in an earlier event, perhaps a subflare which began at 2037 UT. Our X-ray data do not indicate any major enhancement in the interval 2042-2129 UT, however, and the cm-A burst at 2036 UT was very minor. B. JUNE 1, 1967 Data relating to the phenomena of June 1 are shown in Figure 3. Superposed upon the extended X-ray enhancement which began at 07h00m are many short X-ray bursts which are associated with flares on the visible disk. The X-ray background remains above normal until at least 15hOOm. Sacramento Peak reported loops at the west limb, in the position of plage #8818, at the beginning of observations at 12h14m. The McMath-Hulbert flarepatrol films record their presence at the beginning of observations at 11h03m We have no data which provide the time of first visibility on the sun. Evidence for an initiatory flare has been discussed by Teske (1970). It is likely that the X-ray rise after 07ho0m is associated with the flare and the 87

radio burst. Further, the maximum X-ray flux recorded at around 09 00 may be to a slight extent contaminated by subflaring activity near that time. It is unlikely, though, that the maximum seen in soft X-rays represents only the X-rays of the initiatory flare, because it occurs 110 min after the peak of the cm-A burst. It is unfortunate that swept-frequency observations at meter-A were not being made around the time of the flare. C. MAY 21, 1968 "Loops" were reported by Sacramento Peak Observatory on the northeast limb beginning at 20h50m, eventually extending to 0.08 (h - 6 x 104km) above the limb. These thus rank with average loop heights recorded by Bruzek (1964). Our flare patrol film, which ceased at 20h56m, indicate structures that resemble a great loop system as early a.s 20h41m. The assembled data for this event are in Figure 4. A subflare on the northeast limb may be associated with the formation of the loop system. If so, its importance rating is perhaps incorrect. A Type I.I burst following the flare is suggestive of a significant magnitude for the event, while the cm-A burst amplitudes (a 20 flux units) would argue otherwise. As on October 30 and 51, 1967, no Type IV radiation was reported. It is likely that the X-ray enhancement between 20 20 and 21 00 is primarily a burst component associated with the subflare. The burst is more prolonged than is usual for a subflare, although the rate of decay of the burst is essentially flare-like. The low level of enhancement after 22h00m demonstrates that this loop prominence system was not as productive of X-radiation as were the others previously discussed. The Ha data available to us do not indicate that the system was morphologically different from other major loop systems, although these data are of poor quality. Do X-RAY ENERGIES OF THESE EVENTS We have estimated the peak emission rates and time-integrated soft X-ray energy of three of the events (Table III). That given for the June 1, 1967, loops is surely an overestimate because of the detection of some flare burst X-rays but is not too large by more than a factor of two. Thomas (1970) has analyzed the response of the Michigan ion chamber to various spectral distributions-both line and continuous-and concluded that the relative errors of quoted fluxes do not exceed 30% for the spectral distributions that may be reasonably expected even if time-variations of spectrum occur. He has also concluded that systematic sources of error (calibration, primarily) do not lead to errors in quoted energies exceeding a factor of three, and that the true systematic errors may be less than that. 88

TABLE III SOFT X-RAY EMISSION RATES, ENERGIES, RISE, AND DECAY TIMES FOR THREE MAJOR LOOP SYSTEMS e-folding times X-Ray Peak Flux Date X-Ray Peak Flux JAE(8,12) dt(ergs) Rise Decay at Sun (ergs ~se )(Minutes) (Minutes) 6/ 1/67 6.2 x 1025 1. x 10 10/30/67 4.2 x 1025 6 x 1019 l 216m 10/31/67 5.7 x 10 5 6 x 109 119 243 It is surprising to find three separate events which produce such similar peak rates and total X-ray energies. Energy estimates cannot reliably be made for the May 21, 1968, loops. However, the peak rate of X-radiation and total X-ray energy for the loops on that date are very much smaller than those given in Table II. The following discussion (Section III) will consider the loop systems of October 50 and 31 at length. These were the largest loops that were identified optically, and were among the strongest soft X-ray emitters in Table I. They may therefore not be typical of loop systems generally; the loops of May 21, 1968 may also exemplify a morphologically similar but less energetic phenomenon. III. Disc-ussion Because the Michigan ion chamber responds to a broad band of wavelength principally between 8 and 12 A (see Thomas, 1970) it is not possible to utilize the observations directly in a deductive physical discussion. To do so would require prior information as to emission measure and temperature. Comparison of our data with solar X-ray data at 7,7 - 210 keV energies obtained in flare events by Peterson and Hudson* indicates that in flares the flux levels observed by us are almost entirely line or free-bound emissions. This is deduced from the fact that an extrapolation of the bremsstrahlung spectrum near 7.7 keV into our band-pass fails to explain our flux levels by factors of from 10 to 100. The complex interplay of variations of emission lines and continua during the course of a transient phenomenon makes it extremely difficult to extract any but rudimentary information. However, Thomas (1970) has pointed out that the relative flux levels measured are not significantly affected by spectral variations. We thank Drs. L. E. Peterson and H. S.''Hudson of the University of California at La Jolla for an opportunity to inspect their data. 89

During the course of the October 30 and 31 loops, the cm-A radio emission was weak, non-impulsive and probably thermal in origin. Thus our X-ray fluxes probably represent line emission and some free-bound continua, for which the flux levels are given by E(8,12) c / N 2 f(T ) dV - N e V f(T ) (1) V where we assume an isothermal-isotropic emitting volume. f(Te) is a function of temperature appropriate to the thermal emission mechanism. The X-ray flux curves observed in October, 1967 are not directly associated with the inferred flare events, but principally with the loops. We ask whether the loops might be the result of a compression that is quite separate from a flare process. In considering the Olson-Lykoudis compression model, then, we may write (N V) = (Ntotal ) and 2 Ntotal E(8,12) c f f(Te) (2) since no material is to be added to the magnetic field lines. During the compression stage, if it occurs, the decrease in volume, V, may possibly act to provide a temporarily increased rate of line and continuum emission. There are two difficulties with the compression model, however, beyond those already cited. The first difficulty is that the Olson-Lykoudis model predicts that the highest temperatures will be achieved while the compression is still in progress, and Te will have declined significantly before the compression is completed. Thus, if indeed sufficient X-radiation is emitted, X-ray maximum will occur prior to the visible appearance of Ha loops in the system. This was not the case on October 30, 1967. A second difficulty is inherent in the model calculations: an expression derived by Oriall and Zirker (1961) was used to specify the volume emission coefficient; comparison of this expression with the calculations of Cox and Tucker (1969) shows it to be too small by about a factor of four in the relevant temperature range. Thus the predicted timescale for development of visible loops in the compression-condensation model must be shortened well below the observed time-scale for the decay of the soft X-ray emission. We must therefore reject the model as incompatible with the observations. In conjecturing that major loop systems are visible evidence for the storage of fast particles, Jeffries and Orrall (1965a,b) suggested that injection of the particles onto magnetic field lines above the flare site leads to increased density near the tops of the field loops, where the keV protons are thermalized. As density and temperature rise the region will expand rapidly 90

and, cooling, produce the visible Ha loops. It is of interest to ask whether protons are injected on a short time-scale or over a long interval of time. Jeffries and Orrall (1965b) suggested that a continued injection may sometimes occur. We assume first that all of the loops of a system are a. consequence of a single event taking place on a short time-scale and that the protons are directly injected into the magnetic field loops. Reasons have been cited for expecting expression (1) to be applicable to the observations under discussion, at a time when the plasma is thermalizedo If there is a single injection event, we may use expression (2) in discussing a single loop subsequent to the thermalization stage and explosion at the top of the loop. At the time of the explosion, Ntotal will be constant, V increasing and Te diminishing, leading to a decline of soft X-ray emission from that loop. On magnetic field lines threading higher altitudes, however, the plasma will not yet be thermalized, since thermalization time increases with decreasing mean ambient density. While we do not know the initial energies of the injected protons, the calculations of Jeffries and Orrall (1965a) indicate that for reasonable energies, protons injected into even very high loops within the permanent coronal condensation should be thermalized within 10 min. If, after thermalization and the conjectured explosion the plasma may be characterized by Te ~ 10 ~K and Ne - 1011 cm-3, we may calculate the time for the gas to cool to a temperature at which the Ha line becomes visible, using Cox and Tucker's (1969) emission coefficients. Assuming that the rate of change of internal energy of the gas is approximately dT dE e -3 -1 dt 3.5 NHk d ergs cm sec over the relevant temperature range (H and He fully ionized), we find that Te will decline from 107 OK to 5 x 104 OK in 1.5 x 105 seconds by radiative processes alone. Electron conduction may also be an important cooling process. A comparison of the total time for thermalization and radiative cooling in a single loop ( 35 mmin for high loops to first become visible under the assumptions made) with the soft X-ray decay time-scale for the entire system (~ 220 - 240 min) does not support a single-injection scheme. On the other hand, if the plasma is driven to Te 10 ~K during thermalization of the protons, the time-scale for radiative cooling to 5 x 10 ~K is 1.8 x 10,1/Ne seconds. The observed time-scale for X-ray decline then requires Ne - 1.3 x 1011 cm53. While this number density satisfies the optical observations (Jeffries and Orrall, 1965a), we can show that the single-injection model, if it is valid, would not produce such a high kinetic temperature. In the single-injection model, all protons will be thermalized with ~ 10 min, and therefore before radiative losses have caused a significant decline in 91

Te, if Te is 10 ~K. Kleczek (1964) has measured the total volumes of loop prominences V > 2 x 1027 cm3. Thus, an emission measure N2 V > 2 x 1049 cm3 may be estimated from the requirement Ne = 1 x 101 cm3. If Te rises to 10~ ~K, the result will be an enormous flux of hard thermal bremsstrahlung radiation, which is not observed. The single-injection model then cannot be invoked to satisfy the soft X-ray observations if Ne 1011 cm-3 in the loops. We have the alternatives of (i) supposing Ne < 1011 cm-3 or (ii) invoking an injection on a longer timescale. As to the former alternative, the X-ray decay time may be satisfied for Te = 107 OK and Ne = 1.2 x 1010 cm- at the tops of the loops following thermalizationo There still remains, however, the difficulty of the long time interval over which successively higher loops appear in the prominence system. No single loop of a major system is seen to enlarge or to rise in altitude; rather the loops seem to appear at successively higher levels. Several hours after the initiating flare, loops are still appearing at high altitudes, but at a much reduced rate (Bruzek, 1964). Undoubtedly the very late appearance of higher loops is associated with longer thermalization times at the higher altitudes and with a longer radiative cooling time imposed by the lower densities. We suggest that it must also be associated with a later injection of energetic protons onto the highest loops. The protons may be stored for long periods on magnetic field lines that rise beyond the inner corona (Jeffries and Orrall, 1965b) and delivered to lower field lines to produce the loop system. Our X-ray fluxes, which depend upon NeVV, probably protray the delivery of the particles onto the prominence loops rather than the storage reservoir itself. It is not clear to us, however, why particles stored on field lines rising to high levels should be delivered to the lowest-lying loops of the prominence system first. An alternative to the storage model is the hypothesis that protons continue to be accelerated within the initiatory flare during the formation of the loop system. Under this hypothesis, our X-ray flux curves then show a slow initial development and slower decay of the secondary acceleration mechanism. IV. Summary Major systems of loop prominences appear to be productive of soft Xradiation on a long time-scale. In at least one such system, for which our identification is not entirely clear-cut, the X-ray emission was however quite weak. 92

We have compared observations of X-rays from two systems of loops (October 30 and 31, 1967) with two models for production of the loops. The data are not compatible with a compression-condensation model. The data are however compatible with the single-injection model provided (i) Ne in the loops is - 1010cm-3 an order of magnitude lower than the value derived from optical spectra. The particle-injection model may also be invoked, on the other hand, provided (ii) the delivery of protons into the prominence system takes place on a time-scale of hours. Our data cannot distinguish between these alternatives. 95

BIBLIOGRAPHY Bruzek, A. 1964. Ap. J. 140, 746. Cox, D. P. and W. H. Tucker. 1969. Ap. J. 157, 1166. Dodson, H. W. 1961. Proc. Nat. Acad. Sci. 47, 901. Hyder, C. Private communication. Jeffries, J. T. and F. Q. Orrall. 1965a. Ap. J. 141, 505. _. 1965b. Ap. J. 141, 519. Kleczek, J. 1964. AAS-NASA Symposium on the Physics of Solar Flares, NASA SP-50. Olson, C. A. and P. S. Lykoudis. 1967. Ap. J. 150, 303. Orrall, F. Q. and J. Zirker. 1961. Ap. J. 134, 72. Teske, R. G. 1969. Solar Phys. 6, 193. Teske, R. G. 1970. Paper V of this report. Thomas, R. J. 1970. Dissertation, University of Michigan. 94

FIGURE CAPTIONS Figure 1. In this summary of solar data for October 30 and 31, 1967, the top panels depict times of beginning, maximum and ending of reported flares and the reported times of swept-frequency radio spectrum events. Below these is shown the soft X-ray flux, which is given as the energy flux integrated over the wavelength interval 8 to 12 A as measured at the earth. In the lower half of the figure, singlefrequency radio bursts are schematized according to the author's interpretation of data published in the Quarterly Bulletin of Solar Activity. Radio flux density is measured in units of 10 watts m Hz. The observed times of loop prominence activity are shown just above the X-ray record. Figure 2. a. Sacramento Peak Observatory film of west limb loops, 21 42m53s UT October 31, 1967. H-alpha. b. Sacramento Peak Observatory film of west limb loops, 2h34m20s UT October 31, 1967. H-alpha. c. Lockheed Observatory flare-patrol film (H-alpha) taken at 20hO8m UT October 30, 1967. Figure 3. Solar data summary for June 1, 1967. The top panels depict flare occurrence and reported swept-frequency radio spectrum events. Below these we schematically show reported cm-A and dcm-A singlefrequency bursts. At the bottom is the soft X-ray flux record. Loop prominence occurrence is shown just above the X-ray record. Figure 4. Solar data summary for May 21, 1968. Flares, swept-frequency radio spectrum events, single-frequency bursts and X-ray fluxes are shown. Loop prominence occurrence is shown just above the X-ray flux record. 95

sI A — h- A ILI —-_A_ _III —- -I II __ __AA RADIO SPECTRUM TYPE 1Tr IlllI I II I I I II LOOPS LOOPS I. ----- > I ----- > -2 -I E(8,12)ergs cm sec 0.025 0.020 0.015 7 k r o~~ o1 \ -~_ 0.010 J N 117 K 0.005 - J C I I I'I 600 - 408 Mhz 400 - 200 - AA ________________ ________________ 01 A A 50 8800 Mhz 0~~~ 50 F 2800 Mhz 0[ 50 - 2700 Mhz [ 0 50. 1415 Mhz A0 A1 I I,,,. I,,. I,. I.,, I I 12ho0m 18 hoom ooWhom 06ho0m 12ho0m 18hoom 00ooho0 OCTOBER 30, 1967 OCTOBER 31, 1967 Figure 1 96

Ell.............................................................................................................................................................................................................................................................................................................................................................................................................................................................................................'MEN........... NOW MEN'..........

1 I.... I'''' I..... I'.'.'' I FLARE IMP. 2. S J-o... ~ J-4-i L IL- I,ILI I. p-4 I r RADIO T - SPECTRUM TYPE J I II I II -----— } —----— I"' 1 I E TYPE L No Data FLUX 80 >3000 Mhz FLUX 80 (485) 1000 - 3000 Mhz LOOPS,nr E(8,12) ergs cm-2 sec-' 0.03 - 0.02 0.01 0 I','' I, I,,I I O0h 06h 12h 18 24h JUNE I, 1967 Figure 3 98

I I' I W w |' FLARE IMP. S b C ^'^ ^ ^ C C ~ hi i ^ -1 I -,,, > ~ 4 1 4 t RADIO T SPECTRUM TYPE m TIYP I I I II i I i m FLUX 80 >3000 Mhz FLUX 80 1000 - 3000 Mhz LOOPS E(8,12)ergs cm-2 sec'' 0.02 0.0O1 -, fJ --- _,_ n,_ orIo - -\. aI 12h 18h 24h 06h MAY 21, 1968 MAY 22, 1968 Figure 4 99

SOFT SOLAR X-RAYS AND SOLAR ACTIVITY V. Some Evidence for the Altitude of X-Ray Source Volumes in Solar Flares Introduction Currently, solar flare observations extending from the X-ray region of the spectrum into the longer radio wavelengths are provding information on flare structure and the physical processes within them. Interpretation of the data has left some questions open, however. The altitude of the flare-associated hard X-ray source has been placed by some authors at solar atmospheric levels where Ne ~ 109cm-3 (Hudson, Peterson and Schwartz, 1969). Arnoldy, et al. (1968), have suggested that the microwave and hard X-ray source volumes are coextensive, and also have argued that the radio data alone may be consistent with Ne - 104cm-3. Takakura and Kai (1966) present evidence in favor of a cm-k and X-ray source volume at levels where Ne 5 x 1010cm-3, while Kiepenheuer (1964) suggested that flares occur at a level at which Ne - 01-2xlOlOcm-3. Zirin, et al. (1969) have discussed an event in which the hard X-ray source appeared to be co-spatial with the MI source, while Vaiana et al. (1968), interpret their June 8, 1968, flare photograph as showing that the HI and X-ray source volumes are not coextensive Estimates of the emission measure, fNS dV, for X-ray source regions vary from l047cm-3 (Hudson, et al., 1969) to a few times 1048cm3 (Culhane and Phillips, 1970). These are in general agreement with the sizes of cm-. radio-emitting sources which are deduced from interferometry (Kundu, 1965), only if electron densities near 109cm-3 are invoked. There thus is as yet no clearly general agreement as to the positions and spatial organization of X-ray, cm-X and IH source volumes associated with flares, although these questions are being attacked by experiments-in progress and planned-which observe them with good angular resolution. It is possible, however, to investigate these problems with an experiment having no intrinsic angular resolution, if flare events taking place at and just behind the solar limb are observed. In such cases the spatial resolution is afforded by the sun' limb and by the temporal behavior of radiations that are observed, although the spatial discrimination is, of course, very crude. This contribution represents an attempt to utilize observations of such limb events to supply some information about the relative heights of flare source volumes within the solar atmosphere. The X-ray data which were used were gathered with the Michigan soft X-ray monitor on board the OSO-III (Teske, 1969; Thomas, 1970). This instrument observed the solar soft X-ray component between 8 and 12 A with a time-resolution 101

of 1.7 seconds; the whole disk was monitored without angular resolution. In the data accumulated, there are a number of events of significant X-ray amplitude which associate very poorly or not at all with visible flare phenomena. Almost always at the time of such events there was a flare-rich active center on or just behind the solar limb. We have thereby gained the impression that the OSO-III soft X-ray ion chanber at times viewed X-radiation from elevated source volumes associated with flares which occur slightly beyond the sun's limb (infra-limb flares). This impression is consistent with what is known from studies of SID phenomena (Dodson and Hedeman, 1964), with photographs of elevated X-ray source volumes above non-flaring active centers (Vaiana et al., 1968) and with current concepts of flare morphology. Because our instrument has no angular discrimination, the assignment of an otherwise unidentified X-ray enhancement to an infra-limb flare in a particular active center has great hazards unless the assignment can be based upon reliable criteria other than simply the flare-richness of the region. However, the study of correctlyidentified events can supply rewarding information. Five X-ray events which were observed during 1967 and which likely were associated with infra-limb flares were selected for study. Each is discussed individually below, in order of credibility. The assembled information suggests that the volume in which the "flare impulse" or explosion occurs may, in some flares, possibly be in the vicinity of an altitude of 104km, and that there is a separation of source volumes, with the soft X-ray source lying below the cm-\ source. Observations A. AUGUST 18, 1967, 02h23mUT Several major X-ray bursts accompanied the east limb passage of McMath plage #8942. A very extensive and bright plage containing large and magneticallycomplex spot groups, it appears to have been the amalgamation of several flarerich active centers from the previous rotation. During its disk passage, plage #8942, like its parents, dominated solar activity. The two greatest X-ray events observed by OSO-III on August 18, 1967, have the appearance of homologous events.* The second of these accompanied an imp in flare on the visible side of the limb, beginning at 21h31mUT. The first X-ray event began at 0223 UT, and we have inferred it to have been associated with an infra-limb flare in the same region. In Figure 1 we present the X-ray and radio observations for the time of the suspected infra-limb flare. The radio data were kindly communicated to us by Dr. S. Enome at Toyokawa. During the important time interval, the Haleakala photographic flare patrol and the Ikomasan visual patrol were in operation. *This was initially pointed out to us by Dr. Hugh Hudson. 102

Both observatories reported a subflare at the east limb. Dr. John Jeffries has very kindly made available to us copies of the Haleakala filtroheliograms. These show the appearance of a small flare (projected area - 100 millionths of the disk) from behind the east limb at 0239 UT, its maximum at 0248 UT (see Figure 2) and its disappearance at 0306 UT. Although other flaring activity was going on in plage #8940 on the visible disk around the times of interest to us, it is difficult to associate the major radio and X-ray bursts that were observed with that visible activity if we use the timing and relative flux amplitude criteria that have been established for the soft X-rays and cm-k bursts (Teske and Thomas, 1969). On the other hand the same criteria are also violated by the very late appearance of the limb flare and by its small area, but if we postulate that the visible limb phenomenon was a part of a larger infra-limb flare, the violations may be explained. By intercomparing data from the Michigan soft X-ray photometer and the University of California hard X-ray counter (which also observed the event from OSO-III) we have erected criteria for judging the consistency of behavior of flare events. By these criteria for visible flares, the soft X-rays, hard X-rays and cm-k flux for this flare are all in general agreement with one another as regards relative starting times, times of maxima and relative peak fluxes (or spectrum). Moreover, the event showed decided impulsive characteristics at those wavelengths that were observed. However, no swept-frequency m-k events (Types II, III, or IV) were reported near the times of interest, although Sydney was observing. In our interpretation of this event, both the region of the "flare impulse" or explosion and most or all of the X-ray and radio source volumes were above the solar limb and were visible at the earth. The magnitude of these emissions together with the late appearance, small projected area and feeble brightness of the HA aspect suggests that only a part of a larger, lower-lying optical source was seen at visual wavelengths. We have estimated the longitude of the flare to have been 9~ (+3o,-6~) behind the solar limb by plotting the longitudes of subsequent flares, fitting them to a line of slope equal to the rate of solar rotation at the flares' mean latitude and by extrapolating the line back to the time of radio burst onset. Thus, heights down to within 104 km above the flare were being observed. The visible part of the flare may have extended to - 2.0 x 104 km above its base. B. JUNE 1, 1967, 07hOOm UT McMath plage region #8818 transited the disk during the latter half of May, 1967. One of the most active regions of 1967, it produced two PCA events (May 23 and May 28) and many other major flares. By May 30, flaring was most pronounced in the western part of the region so that during its west limb passage on May 31, June 1 and 2, the most active portions set behind the limb first. 103

On June 1 a strong soft X-ray enhancement began at 0700 UT, near the time of a major burst on 2000 MHz which was not identifiable with a visible HX flare (see Figure 3). The X-ray background remained high until at least 1500 UT. A loop prominence system was subsequently observed on the west limb in the position angle of plage #8818 (Teske, 1970). The loop system was well developed by 1103 UT when cinematographic observations were commenced at McMath-Hulbert Observatory. We have inferred from these observations that an infra-limb flare probably occurred in plage #8818 behind the west limb. The estimated flare longitude is W98~ (~6~), so that heights down to - 104 km above the flare were being observed. Although the radio burst on 2000 MHz was strong (485 flux units) and impulsive, the radiation at higher frequencies was weaker and less impulsive, implying an unusual radio-frequency spectrum. The accompanying soft X-ray burst, between 0700-0750 UT, was characterized by a slow rise having a less steep slope than is usually found in association with major cm-X and dcm-K bursts. The X-ray maximum occurred > 20 min after the 2000 MHz maximum, very long in comparison with bursts accompanying visible flares (Teske and Thomas, 1969). Our data indicate that the soft X-ray maximum may have occurred near 0900 UT, but this maximum is probably contaminated to an unknown extent by X-rays from a subflare and likely contains a component associated with the loop prominence system as well (Teske, 1970). When we take this into account, however, we find that the soft X-ray amplitude associated with the radio burst is not in disaccord with the 3100 MHz burst amplitude (Teske and Thomas, 1969). Our interpretation of these observations, the relatively slow X-ray rise and its delayed maximum which eventually reached a flux level appropriate to the cm-X amplitude, is that the 2000 MHz source lay at atmospheric levels above the soft X-ray source, which then rose or expanded until a major fraction, at least, was visible from earth. The impulsive volume of the 2000 MHz source lay at altitudes t 104 km, a conclusion that is consistent with the electron density model of the solar atmosphere. The inferred expansion or upward propagation of the soft X-ray source volume may be connected with the subsequent formation of the system of loop prominences. C. OCTOBER 30, 1967 During its transit of the western side of the solar disk, McMath plage region #9034 developed into a richly-flaring active center; the incumbent spot group was classified as 8 on October 26 (and was not classified again until very close to the limb on October 29, when it was described as P). The increasing pace of visible flaring activity after October 25 was culminated by a flare of importance 2 on October 29 which was associated with a major cm-. impulsive burst followed by a strong, long-enduring post-burst phase and reported loop prominences. Two long-lived soft X-ray enhancements were observed on October 30 and 31, during and after west limb passage of the region. Both events were accompanied by the appearance of major loop prominence systems (Teske, 1970). 104

The soft X-ray, radio and optical data for the two days are summarized in Figure 4. In the figure, the decline in X-ray background is caused by the disappearance of region #9034 behind the sun's limb. The shorter X-ray bursts accompanied flares in other active centers on the visible disk. We first discuss the October 30 event. The imp in flare in #9034 on October 30 (1317-1320-1407 UT) apparently was unaccompanied by a strong soft X-ray burst or radio burst. (In flares, generally, a reportable cm-k burst is accompanied by a detectable soft X-ray burst, although a weak X-ray burst is not necessarily accompanied by a reportable cm-k burst. In the mode of operation of our photometer during October 30 and 31, X-ray bursts of small amplitude -AE(8,12)'0.001 ergs cm-2sec-l- would not have been recorded. Such small X-ray bursts are not usually accompanied by reportable cm-) events.) The McMath flare patrol films start at 1401 UT and show an active prominence at the limb position of #9034. Activity in the region continued sporadically during the day on our films with small surges, the development of a small eruptive prominence and muted brightenings at the limb. None of these suggest themselves as obvious sources for the prolonged slow rise in X-ray flux that began at about 1600 UT. An ADF reported by Sacramento Peak Observatory (N24 E42) between 1654 UT and 2400 UT is in the wrong disk position to have produced the X-rays, which were observed to come from the west limb by an angularly-resolving experiment on board OSO-IV (Kreiger, 1969). However, Sacramento Peak reported possible very faint loops at the position of #9034 between 1654E UT and 1746U UT and bright loops seen edgewise between 2019E UT and 2400D UT. The loops were not present on exposures made at 1813 UT. The system can be seen faintly on McMath flare patrol films beginning at 1850 UT and developed slowly until the end of observations at 1956 UT. Dr. Charles Hyder has kindly re-examined the Sacramento Peak films and reports that the loop system was faint (~0.1 chromospheric intensity in Ha) when first seen when clouds cleared after 2019 UT, and that the loops extended to an altitude of 1.0 x 105 km by 2201 UT reaching,l.2 x 105 km by 2315 UT (Hyder, 1969). Lockheed Observatory reported a flare in #90534 at (1959-2008-2035 UT). Figure 5c reproduces the Lockheed Hm frame exposed at 2008 UT showing the reported flare extending from the northwest limb in the disk field of the photograph and the feature which is identified as the loop prominence system in the limb field of the photograph. Mr. Samuel Miller very kindly made the photograph available to us. The flare in Figure 5c occurred after the loop prominence system was already well-developed. Loop prominences are usually associated with the late phases of flares of imp 2 and 3 (Bruzek, 1964) especially with those that produce Type IV radiation (Dodson, 1961), although Bruzek's list shows that loop systems at the limb are seldom accompanied by Type IV (Bruzek, 1964). The absence of Type IV emission in the present case (see below) does not strongly militate against an inference that a major flare occurred just behind the limb, while the major system of loop prominences would appear to be sufficient evidence for this inference. No definitive optical evidence, other than the loop system, can be cited. 105

At 1725 UT a low bright mound was seen on our Hx flare patrol films in the limb position of plage #9034. It became relatively prominent at 1737 UT, reached a maximum height and brightness at 1759 UT when two small surges developed, and had subsided by 1850 UT when the first evidence for the loops could be seen. We do not regard this evidence as definitive. The major cm-k burst began at 1737+ UT and reached a peak (-50 flux units on 2800 MHz) at -1754 UT. During this interval-1724 to 1824 UT-the X-rays displayed a slow monotonic rise. Such a relationship between the cm-k burst component and soft X-rays is not at all like that developed for visible flares (Teske and Thomas, 1969). The radio bursts and X-rays thus cannot easily be associated with the visible subflare that occurred in plage #9041 near these times (1707-1724-1754 UT). Type III bursts were occurring sporadically in the interval 1429-1530 UT, somewhat earlier than the times of interest, Type I and continuum were reported throughout the interval 1330-2340 UT. Type IV was not reported, and the Fort Davis records were generously re-examined by Dr. Maxwell, who again reported no evidence of any kind for Type IV (Maxwell, 1969). Thus there is no indirect evidence for an acceleration event that would have produced an impulsive component. It seems to us likely that the bulk of the soft X-radiation observed in this event and that on October 31 was associated with the loop prominenece systems (Teske, 1970). We note that the cm-k burst, if it had accompanied a flare on the visible disk, would be expected to be associated with a soft X-ray burst of amplitude AE(8,12)-0.03 - 0.04 ergs cm-2sec-l, and in reasonable timeassociation (Teske and Thomas, 1969). Because no X-ray burst was seen with the cm-k burst, the X-ray source volume associated with the cm-K source may have been at atmospheric levels below the cm-k source and partially or wholly hidden by the sun's limb. In summary, we infer on the evidence of the loop prominences and the cm-k event the occurrence of a major infra-limb flare on October 30, beginning prior to 1850 UT and perhaps as early as 1737 UT, when the cm-k burst began. The absence of a sharp increase in soft X-ray flux at the time of the 1757 UT radio burst suggests a separation of source volumes, with the soft X-ray source volume being the lower. We have estimated, again by plotting flare longitudes and extrapolating, that the inferred flare occurred 9~ (+2~, -7~) behind the limb, giving a sagittal height of -104 km. If we speculate that the loop prominences are evidence for an acceleration event, and that an impulsive phase must have taken place, then the impulsive volume lay below the 104 km level. The absence of Type II and Type IV emissions are arrayed against this speculation, however, although it is unusual for loops at the limb to be associated with Type IV (Bruzek, 1964)o Yet we note the previously-discussed event of August 18, 1967, had an impulsive component without giving rise to Type II, III, or IV radiations. 106

D. OCTOBER 31, 1967 A loop prominence system was observed optically on October 31 at the northwest limb position of plage #9034, and a slow X-ray event like that of the previous day was again recorded in time-association with the optical event (Figure 4). We infer from the occurrence of the loops that a major infra-limb flare took place, beginning prior to 1912 UT, when the first faint tops of the loop system were seen (Hyder, 1969). The loop system extended to 20,000-30,000 km altitude above the limb by 1916 UT and to 40,000 km by 1924 UT, reaching maximum brightness by 1945-1955 UT (Hyder, 1969). We estimate that the inferred flare took place 23~ (+7O, -9~) behind the limb, giving a sagittal height of N6 x 104 km. Thus the loop system probably extended to 105 km above the flare region by 1924 UT. The system, like its predecessor on the day before, thus ranks among the tallest loop systems discussed by Bruzek (1964). Ha filtroheliograms taken at Sacramento Peak have been communicated to us by Dr. Hyder, and are reproduced in Figures 5a and 5b. No Type III or Type IV radiation was observed on October 351 that might be connected with this event. There was no impulsive part to the cm-\ burst. A weak cm- -event thit was in progress near 1500 UT (Figure 4) may be associated with subflaring activity on the visible disk. The long-lived but weak 2700 MHz burst which began just after 1700 UT has no identifiable optical counterpart and may be associated in some way with the inferred flare. Generally, flares accompanied by major systems of loop prominences are major flares of large area which give rise to strong cm-\ radiation and to strong X-radiation. We speculate that on October 31, 1967, such a flare did occur, and that the weak or absent X-radiation and weak cm-X radiation stipulates that the bulk of the X-ray and cm-k sources lay below an atmospheric level of 6 x 105 km. Subsequent to the main flare onset an expansion or outward propagation of the soft X-ray source volume may have taken place that shifted the soft X-ray source to levels above the limb that made it visible from the earth, Alternatively we may speculate that the active volume incorporating the loop prominences became a source of soft X-radiation after the main flare phase. Since a motion of the X-ray source is unlikely, we prefer the latter interpretation (Teske, 1970) E. JULY 11, 1967, 20h35m UT On July 11, 1967 there occurred a weak, slow X-ray enhancement in association with a small cm-X burst and a Type IV dkm-\ event. The data are summarized in Figure 6. During the time interval of interest no significant activity was visible in our mI flare-patrol records on the disk or on the limb. Because of the intervention of satellite night, we do not know if the X-ray event had an impulsive component. There was no X-ray activity seen during the previous daylight pass, 1917-1948 UT. Although for our observations we have no optical or other data to aid in ascription of the phenomena to an infra-limb flare in any 107

region, the importance of the occurrence of Type IV and soft X-rays in close association without a visible flare has captured our attention. There were then no regions behind the east limb that had been flare-prolific during their previous apparition nor that would be flare-productive after their east limb passage in July. There are, however, two candidates that were then behind the west limb that might have produced a major flare. 1. Plage #8871 (Latitude N28) This region (CMP July 1.3) with its incumbent large a spot rounded the northwest limb on July 8. While it was visible on the disk, only a few subflares were seen prior to July 5, when its pace of flaring picked up and continued moderately high until west limb passage. These flares were of imp 1 and 1-. After July 5 the spot was bifurcated and diminished, although the trailing part of the plage grew in brightness. The flares took place preferentially in a part of the plage preceding the a spot which rounded the limb on the 7th and within the newlydeveloping trailing part of the plage which rounded the limb on the 8th. By July 7, all flares were occurring in the trailing region, an observation that may however be due to the extreme limbward position of the large Ca spot. In its next apparition at the east limb on July 21 the region was bright and extensive, incorporating a large and magnetically complex spot group. This daughter plage, designated #8905, dominated solar activity during late July and early August. At its subsequent east limb appearance on August 18, the region was designated #8942 and gave rise to the first event studied in this communication. If a flare in this region did act to produce the observed Type IV and X-radiation, it must have taken place roughly 40~ behind the west limb, and the emitting structures have to have risen - 2.5 x 105 km to become at least partially visible to the earth. While this rather extreme, such structures have been observed at radio frequencies (Kundu, 1965). Selection of this region to explain the observations then requires that the observed Type IV and X-rays arise in a volume containing relativistic electrons that extends a significant fraction of a solar radius out into the corona. It then becomes difficult to understand the short duration of the cm-k burst relative to the durability of the other radiations. 2o Plage #8875 (Latitude S22) The sunspot group in plage #8875 (CMP July 4.5) underwent rapid growth on July 1 and 2. On the 4th and 5th considerable flaring activity was observed in it, with flares all of imp < o1 On the 9th an imp lb flare was reported, while on July 11 one cinematographic flare patrol station (Cape) reported an imp 2n flare on the limb at 1027 UT. We have no X-ray data for that time. The spot group, reported 6 on the 5th and 7th, had simplified to a by west limb passage on the 10th of July. 108

Ascription of the observed Type IV and soft X-rays to an infra-limb flare in this region poses less height difficulties than does ascription to plage #8871. Extrapolated flare longitudes indicate it should have taken place 11~ (+2~) behind the limb; sagittal heights down to 2 x 104 km were being observed. That no visible flare was observed is consistent with the observations of August 18, 1967 and October 30, 1967, already discussed. Further, the relative amplitudes of cm-X burst and soft X-ray burst are generally in accord with the relation established for visible flares, although the relative times of maxima are in violation of the established relative timing for maxima. The observation of Type IV radiation leads us to believe that a particle acceleration event took place. While our X-ray data do not rule out an impulsive phase, the 2800 MHz data kindly sent to us by Dr. Covington show that there was no impulsive phase at cm-k. Thus, if an acceleration event did occur, it must have taken place at altitudes below 2 x 104 km, if we wish to assign this event to an infralimb flare in place #8875. Summary Soft X-ray, radio-frequency and optical observations have been described for five inferred infra-limb flare events. On August 18, 1967, solar atmospheric levels down to 104 km were observed in the event. A portion of the HX flare was seen above the limb and both the cm-k and X-rays were strongly impulsive. The interrelations among the burst timing and amplitudes were such as to suggest that most or all of the source volumes were visible to the earth. In the event of June 1, 1967, atmospheric levels down to 104 km were again observed. While the 2000 MHz burst was strongly impulsive, the soft X-rays were not, and we have inferred a separation of source volumes with the cm-k source lying at or above 104 km altitude, and the initial X-ray source below this. During the event, the X-ray source appears to have expanded to higher levels, a phenomenon that was probably associated with formation of a system of loop prominences. Atmospheric levels down to - 104 km were again observed on October 50, 1967. The cm-k and soft X-.ray bursts were not strongly impulsive, although the occurrence of loop prominences may require that an acceleration event did occur. The data again suggest a separation of source volumes, with the greater part of the soft X-rays associated with the main flare phase being generated below the sagittal level while at least a part of the cm-k source was located above that altitude. The slow development of the soft X-ray enhancement was associated with formation of loop prominences. 109

On October 31, 1967, atmospheric levels below 6 x 104 km were unobservable. The data suggest that during the inferred main flare phase both the soft X-ray and cm-k sources lay below that altitude. As on October 30, a slow development of soft X-radiation was associated with a major loop prominence system. The July 11, 1967 cannot be ascribed reliably to an inferred infra-limb flare in any particular active center. It is not advisable to generalize on basis of this small sample of observed phenomena, especially since an observational selection has strongly biased the choice of events towards those inferred infra-limb flares which give rise to loop prominence systems. It has not escaped our notice, however, that among the first three events discussed, all having roughly the same sagittal height of 104 km, only that event having a clearly identifiable Hx counterpart above the limb displayed impulsive characteristics in both cm-\ and soft X-rays. This comment will apply if we choose to ascribe the July 11, 1967, event to a flare in plage #8875, also. The data suggest that the volume in which the "flare impulse" or explosion occurs certainly lies below 6 x 104 km altitude, and may in some flares lie in the vicinity of h = 104 km, as on October 0, 1967. It is a pleasure to thank the following people who shared their data and/or comments concerning the events discussed: Dr. Arthur Covington, ARO, National Research Council, Canada; Dr. S. Enome, The Research Institute of Atmospherics, Nagoya University; Dr. Charles Hyder, Sacramento Peak Observatory; Dr. John Jeffries and Mrs. Marie McCabe, Institute for Astronomy, University ofHawaii; Dr. A.Kreiger, American Science and Engineering; Dr. Alan Maxwell, Harvard Radio Astronomy Station; Mr. Samuel Miller, Lockheed California Company; Drs. L. E. Peterson and H. S. Hudson, The University of California at La Jolla. 110

BIBLIOGRAPHY Arnoldy, R. L., S. R. Kane, and J. R. Winckler. 1968. Ap, J. 151, 711. Bruzek, A. 1964. Ap. J. 140, 746. Culhane, J. L. and K.J.H. Phillips. 1970. Solar Physics 11, 117. Dodson, H. W. and E. R. Hedeman. 1964. P1. and Sp. Sci. 12, 393. Hudson, H. S., L. E. Peterson, and D. A. Schwartz. 1969. Ap. J. 157, 389. Hyder, C. R. 1969. Personal communication. Kiepenheuer, K. 0. 1964. AAS/NASA Symposium on the Physics of Solar Flares, NASA SP-50. Kreiger, A. 1969. Personal communication. Kundu, M. R. 1965. Solar Radio Astronomy, Interscience Publishers, New York. Maxwell, A. 1969. Personal communication. Takakura, T. and K. Kai. 1966. Pubs. Astron. Soc. Japan 18, 57. Teske, R. G. 1969. Solar Physics 6, 193. Teske, R. G. 1970. Paper IV of this report. Teske, R. G. and R. J. Thomas. 1969. Solar Physics 8, 348. Thomas, R. J. 1970. Dissertation, The University of Michigan. Vaiana, G. S. W. P. Reidy, T. Zehnpfennig, L. Van Speybrock, and R. Giacconi. 1968. Science, 161, 564. Zirin, H., W. Ingham, H. Hudson, and D. McKenzie. 1969. Solar Physics 9, 269. 111

FIGURE CAPTIONS Figure 1. The August 18, 1967, soft X-ray burst-portrayed in the top panelsaturated our OSO-III detector for 8 min. A change in slope on the ascending branch of the X-ray flux curve occurs at the time of the second component of the radio burst. X-ray flux is given as the energy flux integrated over the wavelength interval 8 to 12 A as measured at the earth. —- Radio flux density is measured in units of 10-22 watts m-2XHz1. Figure 2. H-alpha flare patrol frame taken at Haleakala at 02h48m UT August 18, 1967. Solar north is at the top, east to the left. The reported limb subflare may be seen on the east limb. Figure 3. At the top of this summary of solar data for June 1, 1967, we depict the times of beginning, maximum and ending reported for flares. Radio spectrum events are shown below the flare data. On this date, only Type III bursts were reported in the Quarterly Bulletin of Solar Activity (QB). Centimetric and decimetric wavelength radio bursts are shown schematically, according to the author's interpretation of data published in the QB. At bottom is the soft X-ray flux. The loop prominence system was seen in progress at 1103 UT, ending by about 2040 UT, Figure 4. Summary of data for October 30 and 31, 1967. Reported flares and radio spectrum events are shown at the top. Below these is shown the soft X-ray flux record and in the lower half of the figure we schematize single-frequency bursts as reported in the QB. Times of observed loop prominence activity are shown just above the X-ray record. Figure 5. a. Loops in H-alpha at 21h42m53s UT, October 31, 1967. Sacramento Peak film. b. Loops in H-alpha at 22h34m20s UT, October 31, 1967. Sacramento Peak film. c. Flare-patrol frame in H-alpha, 20h08m UT October 30, 1967. Lockheed Observatory film. The feature identified as a system of loop prominences is at the lower left, pointing at the clock image. 112

Figure 6. Data for July 11, 1967, showing reported course of dekameter Type IV burst, soft X-ray flux and 2700 MHz record. No flux scale is provided for the latter, but the scale may be judged from the reported amplitude of the 2800 MHz burst at 2043 UT: 2.6 flux units. The X-ray detector changed its energy sensitivity at 2040 UT and again at 2206 UT. 113

EB (8, 12) ergs cm-2 sec.' S turted u r - S at.12.09-.06 -.03 {FLUX UNITS - AUGUST 18, 1967 FLUX UNITS 500 0400 300 - 9400 Mhz TOYOKAWA, JAPAN 200 - 3750 Mhz 2000 Mhz I00 1000 Mhz t SUBFLARE 0_'',I I A,, I I 0 2ho0m 30m 03h00m 30m U.T. Figure 1 114

- l~~ ~ ~ ~ ~ ~ ~ ~ ~ * Figure 2 115

FLARE IMP. 2I I S L at X hi ^ ^ ^-i ah C H 1 _ C C4 - I RADIO [ SPECTRUM TYPE to'' ": FLUX 80 >3000 Mhz 1 0. ^ A FLUX 80 (485) 1000 - 3000 Mhz = ___ __ LOOPS E(8.12) ergs cm' sec"' J 0.03 - 0.02 - _6L 0.01 - O0h 06h 12h 18h 24h JUNE I, 1967'igure 3 116

2r I FLARE IMR TYPE.EII lI~~~~~I —I II I! II I L-q A - N —---— | ILI| H —----- RADIO T f I SPECTRUM TYPE I[[- Ill~lIIIIl I II I I I II - TYPE LOOPS LOOPS I ----- > I ----- > -2 -I E (8,12)ergs cm sec 0.025 - 0.020 - - 0.010 --''ir jf^ -- r~~~~~~~~~ 0.005- - - oi C ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~I I I I, 600 - 408 Mhz 400 - 20:0- 8800 Mhz 50[2800 Mhz 050 2700 Mhz [ 0 50. - 1415 Mhz 0' AA AA -1 I... I.....II..... I..... I,,,,I,,I Q12hom 18 hoom oo0hom 06h00m 12h00m 18h00m 00oo0m OCTOBER 30, 1967 OCTOBER 31, 1967 Figure 4 117

J,. ~~~~~~~~~~~~~~0'N M X:* >. X ~~ M' ~' >)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~S ~~~~~~~ ~~~~~~~~~~~*>. ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~.... C~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.................. ~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~......5..... I A 8~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.............

Dkm-X Type IT: E (8,12) ergs cmr2 sec-' 0.010 0.008 |L X-Rays 0.006 l 0.004 2700 Mhz,,,.I,,, I 30m 21h00m 30m 22h00m JULY II, 1967 Figure 6

UNIVERSITY OF MICHIGAN IIl 111 111 0358ll111 3 9015 03526 7387