HER X-1 AND CEN X-3 REVISITED

HER X-1 A N D CEN X-3 REVISITED

Riccardo Giacconi

Center for Astrophysics Harvard College Observatory

Smithsonian Astrophysical Observatory Cambridge, Massachusetts 02138

INTRODUCTION

Several galactic x-ray sources, not associated with supernova remnants, have been identified with close binary systems,’ and the data are not in conflict with the view that all remaining sources, a t present still unidentified, will turn out to also be associated with binar ie~.~’ Much of the modeling for these systems has been based on the view that the basic energy source for the x-ray emission is due to accretion of matter from a “normal” star to a compact companion, either a neutron star or a black hole. The observational evidence that supports this view is fur- nished by the rapidity of the fluctuations in many of these systems that compels us to impose strict upper limits on the physical dimensions of the emitting regions and the speeding up, on the average, of the rotational period of the regularly pulsating binaries, Her X-1 and Cen X-3, which is exactly the opposite effect from that expected in a pulsar-like emission process, where the energy emitted in radiation is provided by the loss of rotational energy.

As has been pointed out by some authors, the observational constraints might not yet completely rule out the possibility that the objects in the Her X-l and Cen X-3 systems may be either white dwarfs in which the pulsations may be dbe to nuclear shell burning4 or degenerate differentially rotating white dwarfs emitting pulsed radiation by pulsar-like mechanism^.^ Significant objections have been raised to these models based on the difficulty of obtaining the large observed fluxes6 o r on their apparent inability to account in a natural way for the detailed behavior of the observed emission.

Following the view of several authors,’-”* we have adopted, a t least as a work- ing hypothesis, the view that Cen X-3 and Her X-l are magnetized neutron stars that accrete matter from a “normal” companion.

massive companions would transfer matter through radiation-driven stellar winds, whereas low-mass stars would transfer mass through Roche lobe overflow. Cen X-3 can be considered an example of the first case, whereas Her X-1 can be considered an example of the second case.

X-ray observations of these two systems, which have been analyzed in the last few months, provide a substantial opportunity to test the assumed models. We can distinguish, in the description of these systems, the following different physical regions (FIGURE I ) :

Inside the Alfv6n surface is a region that is defined as the locus of points

As pointed out by some

*A bibliography o f Soviet theoretic research on accretion phenomena can be found in the review by Novikov.*

312

Giacconi: Her X- 1 & Cen X-3 Revisited 31 3

around the neutron star where the magnetic energy density is roughly equal to the kinetic energy density of the infalling material. Within this region, the radius of which is estimated to be of the order of 10' cm for Her X-1, the plasma is con- strained to follow the field lines and will be guided to the star's magnetic poles. The material is corotating with the neutron star, and the details of the accretion process and x-ray production mechanism will affect both the shape and spectral content of the rapid periodic pulsations and the polarization of the emitted radiation. To study this region, one should therefore examine in detail the l .2- and 4.8- sec pulsations from Her X-l and Cen X-3, respectively. Shorter time scales

sec) may provide information about events that occur very close to the surface of the star in the accretion

FIGURE 1. Schematic representation of the rotating neutron star model for pulsating x-ray stars. Both accretion disk and stellar wind cases are shown.

Outside the Alfv6n surface, we can consider the accretion disk. Such disks will occur when the material transferred from the companion star possesses sufficient angular momentum that it cannot fall directly onto the compact object. They have been extensively discussed in connection with the Roche lobe in which case they extend almost to the L1 point. However, such disks may also occur in stellar wind accretion, though over smaller dimensions, and are normally invoked to explain the emission from binary x-ray sources, where the accreting object is thought to be a black hole and the companion a massive star that loses matter by stellar wind. The accreting matter is thought to form a differentially rotating disk comprised of material that gradually spirals inward as viscosity transports its angular momentum outward. For Roche lobe overflow, the disk may extend over regions of the order of 10" cm (Her X-I). Its average density, density distribution, and relative orientation in the system will affect the average observed

3 I4 Annals New York Academy of Sciences

intensity and spectrum of the radiation. If either the neutron star or the disk is precessing, or if substantial changes occur in the matter flow rate, interactions in thesystem may be studied through the examination of the details of the on-off cycles or the existence of absorption features caused by cold matter in the outer regions of the disk. The on-off cycles have characteristic times of the order of several days and weeks; absorption features (dips) may last hours.

At still greater distances from the x-ray source, the flow of the gas from the companion star, and the dependence of density on the distance from the stellar surface, can be studied by considering the details of the transitions in and out of eclipse and absorption features a t particular orbital phases. Typical durations of such effects are in the range of hours. This region can be effectively studied either in x-ray or optical regions. Some of the results obtained by optical observations have been reported by Tananbaum and Hutchings.2 I will therefore limit myself to comments on the x-ray observation that bear on these effects.

At distances of the order of a few times the stellar separation (10'2-10'3 cm), the existence of hot gas with low density (109-10'0 ~ m - ~ ) and extremely long cooling times ( lo4- lo6 sec) could give rise to the low-level emission that is observed in Cen X-3 during eclipses in both on and off states. Typical times for cooling may be as long as a month.

In this short discussion, I will limit myself to describing some of the observational evidence in favor of a stellar wind accretion model for Cen X-3 and some of the recent results on both the spectral content of the rapid pulsation of Cen X-3 and Her X-1 and on their dependence on the on-off cycles.

LARGE-SCALE EFFECTS

Support for the view that accretion in Cen X-3 may be due to stellar winds has until now depended on general evolutionary consideration of massive stars,I2 spectroscopic observations of the Kreminsky star in Hfi and He I 1 4686 with a P Cygny pr0fi1e.l~ and the interpretation of the gradual passage into x-ray eclipse and of the existence of persistent x-ray emission in eclipsed portions of the orbit both in the high and low states.

It is noteworthy that in Her X-1, neither of the last two effects is observed. The transition into eclipse occurs in less than 700 sec. The residual x-ray flux (in the 2-6 keV range) during eclipse or off states is undetectable above background'* and is therefore less than 5 counts/sec.

In Cen X-3, the transition into eclipse is clearly a gradual absorption effect, as demonstrated by the very early results of Uhuru (FIGURE 2) and recently recon- firmed by OAO observation^.'^ The transition is too slow t o be compatible with an exponentially decreasing atmospheric density, too fast t o be compatible with a l / R 2 dependence, and can be explained naturally only by l / R 2 dependence in which ionization effects are taken into account.

compared to Roche lobe overflow ($Iacc = I I M ) also explains in a natural way the persistent x-ray emission of Cen X-3 observed in eclipse. The eclipsed level during the high state of Cen X-3 corresponds to approximately 10% of the maxi-

The inefficient nature of the accretion from a stellar wind (M,,/M =

-t

4

4&21

.000

65

927.

000

8279

7.00

0 13

)933

.000

30.730.000

TIM

E (S

ECO

NDS)

FIG

UR

E

2. A

vera

ge c

ount

ing

rate

fro

m C

en X

-3 in

the

2-6

keV

ene

rgy

rang

e as

a f

unct

ion

of t

ime

obse

rved

on

Jan-

ua

ry 1

1 an

d 12

, 19

71. E

rror

bar

s re

flect

onl

y co

untin

g st

atis

tics.

Mis

sing

dat

a ar

e du

e to

occ

ulta

tion

of t

he s

ourc

e by

the

E

arth

. Dat

a ar

e no

t cor

rect

ed fo

r asp

ect.

F F W

W

ul c

316 Annals New York Academy of Sciences

mum flux (or 15 Uhuru counts/sec vs 160 counts/sec, in the 2-6 keV band). A recent determination by Schreier and Fabbiano of the persistence of radiation in the off state of Cen X-3 reveals a noneclipse value of I I + 1 counts/sec and a value of 5 * 1 counts/sec in eclipse. (Clearly, the intensity in the off state is at times just below the level of detectability during short observations.) The spectral data for eclipsed conditions give a low cutoff energy (E, < 2 keV) and a lower kT ( 5 3 keV) than that which exists for uneclipsed conditions. It should be mentioned jhat the ratio between eclipsed and uneclipsed intensity values in the on and off states is quite different. This fact suggests that the intensity during eclipse is not radiation scattered in the umbra by ionized gas that immediately surrounds the x-ray source, as was suggested by Davidson and Ostriker.' Because the spectrum of the radiation measured by Schreier and Fabbiano is also very different from that of the uneclipsed source (lower kT, low E,,), it is natural to consider that this persistent radiation may come from a shell of optically thin gas at large distances (5-10 times the stellar separation) from the system, thus always unocculted.

It can be estimated that for n, of the order of 10" cm-3 at the neutron star (R - 10" cm) and p - 1/R2, then for R of the order 10l2 cm and at a temperature of IO7-lO8"C, the cooling time is of the order of lo6 sec. The emission then could be due to a spherical shell that surrounds the system. It is clear that this gas could not be heated by the x-rays themselves in a steady state," because L , / n e 2 is too low. We can envisage a situation where gas is heated in close proximity to the source and then expands to large distances in times still short compared to the cooling time. Alternatively, its heating by particles or shocks may turn out to be more effective. Independently from the heating source, the total emission depends on the large amount of matter lost by the system, which is of the order of lo3 times larger than that lost in the Roche lobe overflow mode for the same accreted mass and luminosity. In Roche lobe overflow, matter lost from the system would also tend to be concentrated in the plane of the binary orbit.

More direct and, in my opinion, almost conclusive evidence for the correct- ness of a stellar wind accretion model for Cen X-3 comes from the study of the turnon process in detail. Schreier and Fabbiano discovered that the turnon in Cen X-3 occurred very differently from that in Her X-l (FIGURE 3). Instead of an abrupt increase in intensity, a gradual rise was observed with very peculiar nar- row peaks centered at phase 0.5, which gradually expanded to fill the entire uneclipsed state. Recently, these results have been extended by considering in detail the spectral content of the radiation. The main results of this analysis can be summarized as follows:

Absorption effects by cold gas are observed in the shoulder of the main peak ( E n - 3-4 keV); the radiation in the main peaks exhibits little change in absorption (from E, - 2.5 to En - 2.0) in successive eclipse cycles. The observed effects can very elegantly be explained, at least in a semiquantitative way, by an exten- sion of considerations of Pringle2' and McCray." Pringle considered the effect of the x-ray flux impinging on the stellar wind. He pointed out that the x-ray flux was sufficient to almost fully ionize large regions of the stellar wind atmosphere. He defines an ionized region bounded by a "Stromgren surface" such that the number of recombinations along any radius vector equals the number of ionizing photons. With the parameters suggested by Davidson and Ostriker' in the stellar

al

Wyl

FIG

UR

E

3. C

en X

-3 li

ght

curv

e in

Jul

y 19

72. E

ach

poin

t re

pres

ents

the

ave

rage

int

ensi

ty (

2-6

keV

) ov

er 2

0 se

c, w

ith

back

grou

nd s

ubtr

acte

d, c

orre

cted

for

ele

vatio

n in

the

fie

ld o

f vi

ew o

f th

e co

llim

ator

. T

ypic

al I

u er

ror

bars

inc

lude

cou

nt-

ing

stat

istic

s an

d as

pect

cor

rect

ion

erro

rs. T

he p

redi

cted

ecl

ipse

pha

ses

are

extr

apol

ated

fro

m f

its t

o th

e pu

lsat

ion

phas

e ch

ange

^.'^ T

he g

ap in

cov

erag

e du

ring

Jul

y 27

.5-2

8.5

is d

ue t

o lo

ss o

f po

intin

g. T

he p

oint

s in

dica

ted

by +

sign

s ar

e 2u

up

per

limits

.

w 4

e


wind accretion model used to obtain a Cen X-3 luminosity of lo” erg/sec, namely, a density po at the neutron star of 10llnc ~ m - ~ , an efficiency of wind velocity of - lO’km/sec, he showed that the flux from the x-ray source could reach us with little or no photoelectric absorption over some 280” (through the ionized gas) and suffer considerable absorption in the remaining 80” by the unionized portion of the wind. The point of Pringle’s discussion was to warn us about adopt- ing the size of the occulting region as the size of the companion star.

We have estimated the “Stromgren surfaces” that would result in the event that the density p of the wind was increased from the value po adopted by David- son and Ostriker, of 10’lnc ~ m - ~ , by use of the formulas given by them for mass accretion rate, density dependence on R, and so on, and requiring that L , remain constant.

As can be expected, we find that for increasing density, the cone that contains the ionized gas becomes narrower until finally the “Stromgren surface” completely encloses the x-ray sources. This enclosure occurs at a density p = 40 po. This phenomenon, then, as Pringle had, in fact, suggested, provides a plausible explanation for the offperiods of Cen X-3; namely, the density in the stellar wind becomes so large that it can extinguish the x-ray source at least in the 2-6 keV range or rather shift the radiation to long wavelengths outside our range of observation,

A turnon would then be due to a decrease in density that allows the x-ray radiation to burn through the cold gas and become visible to us (FIGURE 4). The

D

FIGURE 4. A representation of the July 1972 turn-on of Cen X-3. Part A shows the extended low with the source “buried” in the stellar wind. Parts B and C illustrate the appearance and progressive widening of the “spike” near phase 0.5. Part D shows the normal high state, with a wind density at the neutron star of 10” n cm-’.

Giacconi: Her X- 1 & Cen X-3 Revisited

400

> v) 300- W I-

k

E ? W

t 200

LK

100

319

-

-

-

I /

keV

L 200'

3:

CEN X - 3 I I I I I I I I I I I I I I I I \

300' ORBITAL PHASE

FIGURE 5. Light curve of Cen X-3 at two different energies predicted by the model of a turn-on event, as discussed in the text.

condition that L , remain constant while the density decreases is not essential to the argument but may be satisfied in several ways. Either the wind velocity could be decreased ( M a x - v - ~ ) by small factors to compensate for the decrease in density or, as suggested by Tucker, material could be stored in an accretion disk during an onset of a high-density state and then gradually depleted as the turnon progresses. Even more remarkable about this simple explanation is that it apparently satisfactorily describes the detailed appearance and spectral content of the observed turnon event. I f we choose p such that the cone angle equals a width of 90", which is approximately the width of the third eclipse cycle main peak in FIGURE 3, we find that for radiation of different energies, the intensity profiles will be as illustrated in FIGURE 5 . The radiation in the shoulder is severely depleted at energies below about 4 keV. The intensity in the peak is 60% of maximum. The missing 40% will be scattered by the ionized gas and will be redistributed in all directions. Much of this radiation will now reach us, even if in directions outside the ionized cone, because for a given angle, it will traverse much less dense cold regions than will the direct radiation from the star. We estimate that this radiation contributes a constant level as a function of phase of about l0-15% of peak value. The superposition of these effects accounts, a t least qualitatively, for the shape and spectrum of the observed radiation.

The view that radiation-driven stellar winds can, in fact, have large density

Annals New York Academy of Sciences

changes is in accord with theoretic views on the subject23 and does not contradict the scant available observational evidence.

An additional effect observed during turnon is conveniently explained in this model. The 4.8 periodic pulsations were observed to contain a very small fraction of intensity pulsed at the beginning of the onstate; the pulsed fraction mono- tonically increases as the on state progresses (FIGURE 6). We attribute this effect to the increased percentage of unscattered intensity reaching us from the source due to the decrease in optical depth for Thompson scattering in the ionized region. As usual, however, in the study of these systems, the story on transitions is not quite so simple. In FIGURE 7, a transition to a low state is shown. We are immediately struck by the fact that this phenomenon is quite different. Here we find no pronounced intensity or spectral dependence on orbital phase. There is no detectable change in spectrum as the intensity gradually decreases. The source, rather than being “choked,” appears to be “starved.” Because the effects are different, it is interesting to point out how they relate to each other in time. The turnon and turnoff transitions occurred in July 1972 and February 1973, respectively. One

FIGURE 6. Data (2-6 keV) folded modulo the 4.8-sec pulsation period. Several 20-sec passes from each of four separate orbital periods during the July 1972 transition were phased together; each pass was corrected for heliocentric and binary orbit phase variations. The arrow indicates the phase of the fundamental 4.841-sec sine curve fit to aU of the data of the last full orbit. July 22 .

10 (- - -); July 24 (-); July 26 (- . -): July 30 ( * * * *).

1 2 3 4 5 6 7 8 9 10 I_ 4.841 SEC-4

off and one on transition occurred during the intervening months and more may have gone undetected due to discontinuous coverage (FIGURE 8).

Pursuing the choked-starved analogy, we estimate that a decrease in accretion of only a factor of 20 would be sufficient to reduce the intensity of Cen X-3 below the detectability level. This effect could be due to a rise in wind velocity by small factors or by a decrease in density by a factor of 20. The diagram in FIGURE 9 depicts how the source may act as it varies between 40 and 0.5 P O . Although other explanations may be proposed for the turnoff behavior, we find that this model will satisfy simple tests. We should observe different turnon and turnoff transitions from the “choke” and “starve” conditions. A starve-off should be followed by a starve-on and a choke-off by a choke-on. The maximum rate of pulsational period and orbital period changes should be related to the specific high- or low- density condition at the source that corresponds to high or low rates of matter loss. We propose to pursue this investigation utilizing additional Uhuru data. It is clear, however, that continuous observations would be most beneficial, par- ticularly if they could also be performed in the visible and uv spectra, where sub-

Giacconi: Her X-1 & Cen X-3 Revisited

0 0

P

F- 11 n n


t a

8

Giacconi: Her X- 1 8c Cen X-3 Revisited 323

stantial effects should accompany this wide range of wind density and velocity changes.

I believe that it is useful to stress how different the erratic behavior of Cen is from the very regular (though not fully understood) Her X-l turnon and turnoff behavior illustrated i n FIGURES 10 and 1 I .

In the first Figure, we can clearly see the rapid turnon and the absence of intensity dependence on orbital phase. In the second one, the regularity of the turnon that occurs within restricted orbital phases strongly emphasizes the more regular nature of the gas flow that occurs in Her X-I, as would be expected in a Roche lobe overflow, The existence of regularly marching absorption dips also emphasizes

FIGURE 9. The range of p g over which Cen X-3 is observable, based on the model discussed in the test.

the repeatability of the clock mechanism that underlies the Her X-l 35-day cycle. By contrast, we find that absorption dips occur on Cen at about phase 0.75. These dips are rare, which indicates that they are transient phenomena. They do, however, suggest that possibility of an accretion disk in Cen X - 3 . In this view, the dips would be due to density increases in an accretion disk as a result of interaction with an incoming cold gas stream. An alternate explanation is based on absorption by a high-density region in the shock region produced by the interaction of the wind with the neutron star.

I f this general model of Cen X-3 is confirmed by future observations, it will open interesting possibilities for the study of x-ray observations of radiation-

3 24 Annals New York Academy of Sciences

m y*yII. nn

W -

a N W

Y ) -

a -

! I I I I

FIGURE 10. Detailed 2-6 keV intensity observations of Her X-1 during January 1972, March 1972, and July 1972 high states. Each individual sighting of the source is repre- sented by a dot with typical statistical and systematic errors indicated. The curves all show the sharp turn-on, the broad maximum, the gradual decrease over several days, and the presence of the dips discussed in the text. An example of a dip clearly separated from the eclipse can be seen on July 10.9.

driven winds in massive stars, a subject that appears to warrant observational data.

SMALL-SCALE PHENOMENA

Some progress has been made in the study of the short-time periodic pulsations of Cen X-3 and Her X-1. I would like to only briefly mention some of the results to indicate the avenue of research that we are pursuing.

In addition to the investigation reported above by Schreier and Fabbiano on the change of pulsed fraction during Cen X-3 turnon transitions, Ulmer has

2.5

- 1

I I

I I

I I

SOU

RG

E TU

RN

-ON

I

1 I

I 1

I I

- 2.

0 -

DA

YS

zTU

(NO

N T

IME

- N

13

4.8

8

%

- = - -

- L,,,

- 1.

5 -

34.8

5 D

AYS

I0 -

0.

5 -

- - - - i

In

-0.5

-

-1.0

-

/

- -

-2.0

-

I I

I I

I 1

I I

I I

1 I

1 1

FIG

UR

E

11.

Top:

The

dif

fere

nces

in d

ays

betw

een

the

obse

rved

tur

n-on

tim

es o

f th

e H

er X

-1 h

igh

stat

es a

nd w

hat

coul

d be

pre

dict

ed b

y a

con-

st

ant 3

4.88

-day

tria

l per

iod.

The

err

or b

ars

repr

esen

t ab

solu

te l

imits

on

the

time

duri

ng w

hich

the

tur

n-on

occ

urre

d. T

he d

ashe

d lin

es r

epre

sent

co

nsta

nt p

erio

ds a

nd s

how

the

perio

dici

ty i

n tu

rn-o

n ap

pare

nt f

or a

few

cyc

les.

Bof

rorn

: T

he a

bsol

ute

rang

e of

eac

h tu

rn-o

n pl

otte

d as

a f

unct

ion

of o

rbita

l ph

ase.

Pha

se 0

.0 is

the

eclip

se c

ente

r. T

he n

umbe

r as

soci

ated

with

eac

h de

term

inat

ion

refe

rs to

the

absc

issa

of t

he t

op F

igur

e.

wl


examined pulses that occur during an entire 2-day uneclipsed state. We find that the shape of the pulse is generally insensitive to orbital phase. We then superim- pose all available data during this period (which consist of 12 observations of 40 sec each) to increase the available statistical accuracy and examine the pulse shape dependence on energy. No obvious change appears to occur in pulse shape, nor does any shift in phase of the peak with energy occur. However, when we use all the data to obtain spectral fits as a function of phase, we do find a definite dependence of spectral index on pulsational phase (FIGURE 12). These results essentially confirm the previous results by Schreier et al.” The higher temperature (smaller spectral index) associated with the peak of the pulses is equivalent to the statement that the pulse is narrower at higher energies, an effect that has extensively by Tsuruta and Rees,” Gnedin and Sunyaev,26 and

been discussed several others.

FIGURE 12. The “grand average” pulse shape of Cen X-3 and spectral index vs phase derived by fitting a power law to the 2-12 keV flux.

1011 I I I 1 , L , , , , 0 0 2 0 4 0 6 0 8 I 0 PHASE ACROSS PULSE (T= 4.84 sac)

Although perusal of short stretches of data reveals that large pulse-to-pulse spectral variations occur in Cen X-3 on time scales of seconds, the present results indicate that, on the average, pulse shape and spectrum remain relatively constant. The short-term variations are apparently uncoupled to the long-term effects, such as those described during turnon.

Turning now to Her X-1, Joss and others at the Massachusetts Institute of Technology and Jones-Forman and Forman of the Uhuru group have initiated a systematic data analysis program with Uhuru observations to study the variations of pulse shapes and spectral content of the 1.24-sec periodic pulsations as a function of the 1.7-day orbital period and the 35-day on-off cycle.

The results of this analysis for the 35-day cycle correlations obtained at Har-


vard are shown in FIGURE 13. These data were obtained during the 9-day on state in March 1972. Each horizontal group corresponds to pulses averaged over an orbital period modulo the 1.24-sec period, taking into account Doppler shifts due to orbital motions of the x-ray source, satellite orbital motions, and heliocentric corrections. The phase has been preserved throughout this summation. Each horizontal group exhibits the pulse shape in the 2-6 keV integral channel and as a function of different energy intervals. The six horizontal groups are arranged vertically in order of increasing time in the 35-day cycle. Spectral fits were obtained for each of the six groups as a function of five intervals of the 1.24-sec phase.

The striking feature of these results is the obvious variation of pulsed fraction with the 35-day cycle phase. This effect was first noted by Joss and has been confirmed and extended in this work. In particular, the smooth increase of pulse fraction and the gradual appearance of an interpulse are noticeable. It is also interesting that although several investigators had reported that a variety of pulse shapes could be observed to occur on time scales of minute^,'^'*^'*^ no corresponding effect seems to be reflected in the averaged data. I conclude that the changes from single to double pulses occur only on short time scales and are presumably due to effects that take place in the accretion column. Another striking result is that no detectable change in spectral index or cutoff energy is evident throughout the data (with one notable exception), as a function of the 1.24-sec phase or of the 35-day cycle. It would, therefore, appear that if reprocessing of a narrower pulse is taking place, it must occur in the neighborhood of the x-ray star by scattering in ionized matter rather than by absorption and reemission from cold gas.

The noticeable exception mentioned above to the uniformity in spectral index and cutoff energy occurs during the first orbital period after turnon. In this case, a substantially larger cutoff (3 vs 1.5 keV) is observed, and appears to be phase dependent. Perhaps the simplest explanation of this effect is that it is caused by metries of the accretion column, although interactions with matter in the disk could perhaps account for it. This explanation implies that the disk absorptional properties are cyclically changed, possibly due to the beam itself, and then restored within times of the order of the pulsational period.

in any event, the gradual change of pulsed fraction and of the relative intensity of the interpulse to the remainder of the pulse could be explained by geometric effects and would seem to favor explanations based on the precession of the rotating neutron star or, possibly, of the accretion disk. It should, however, be noted that recent results of Joss (which will be reported at the Workshop Session of this meeting) show a slightly different dependence of pulse shape on the phase of the 35-day cycle. Thus, the effects do not appear to be entirely reproducible from cycle to cycle.

We intend to pursue this investigation to examine possible dependence on orbital phase and on effects that occur on shorter time scales. It is clear, however, that we will be severely limited in this endeavor by the limited sensivitity of Uhuru. In several of the problems related to short time domains, it is clear that substantial progress will be achieved only with the much greater sensitivity of the HEAO-A mission.

Giacconi: Her X- 1 & Cen X-3 Revisited 3 29

CONCLUSIONS

I hope this brief description of progress in the x-ray studies of two of the most fascinating objects among galactic x-ray sources has provided some indication of the considerable accomplishments we have made in understanding these systems. I hope I have made equally clear that a considerable state of uncertainty still exists about the very fundamental processes that occur within the Alfvtn surface and that give rise t o the observed phenomena.

ACKNOWLEDGMENTS

I am very grateful to Ethan Schreier, Christine Jones-Forman, William For- man, Me1 Ulrner, Ken Swartz, and Peppi Fabbiano for their help in preparing this talk, which is entirely based on their recent work. I am also grateful to Wally Tucker, Ken Brecher, Mike Lecar, and Paul Joss for their helpful discussions.

REFERENCES

I.

2. 3.

4.

5. 6.

7. 8.

9. 10. I I .

12.

13. 14.

15. 16. 17.

18.

19. 20.

GIACCONI, R. 1974. In Gravitational Radiation and Gravitational Collapse, IAU Symposium No. 64. C. DeWitt-Morette, Ed. : 147. D. Reidel Publishing Co. Dordrecht, The Netherlands.

TANANBAUM, H. D. & J. B. HUTCHINGS. This monograph. GURSKY, H. & E. SCHREIER. In Variable Stars, IAU Symposium No. 67. L. Plaut, Ed.

BLUMENTHAL. G. R., A. CAVALIERE, W. K. ROSE & W. H. TUCKER. 1972. Astrophys.

BRECHER, K. & P. MORRISON. 1973. Astrophys. J..180: L107. KATZ, J. K. & E. E. SALPETER. 1974. Preprint. CRSR 577. Cornell University. Ithaca,

SHKLOVSKY, I. S. 1967. Astrophys. J. Lett. 148: LI. NOVIKOV, 1. D. 1973. Astrophysics and Gravitation.: 317. University of Brussels.

DAVIDSON, A. & J. P. OSTRIKER. 1973. Astrophys. J. 179: 585. PINES, D., C. J. PETHICK & F. K. LAMB. 1973. Ann. N.Y. Acad. Sci. 224: 237. REES, M. J. 1974. In Gravitational Radiation and Gravitational Collapse, IAU Sym-

posium No. 64. C. DeWitt-Morette, Ed. : 194. D. Reidel Publishing Co. Dordrecht, The Netherlands.

V A N DEN HEUVEL, E. P. J. 1974. Preprint. Astronomical Institute. Utrecht, The Netherlands.

DAVIDSON, K. 1973. Nature Phys. Sci. 246: 1. TSURUTA, S. 1974. Preprint. Astronomy Centre, University of Sussex. Falmer,

PRINGLE, J . E. & M. J. REES. 1972. Astron. Astrophys. 21: I . SHAKURA, N. 1. & R. A. SUNYAEV. 1973. Astron. Astrophys. 24: 337. VIDAL, N. V., D. T. WICKRAMASINGHE, B. A. PETERSON & M. S. BESSELL. 1974.

Preprint. Mt. Stromlo and Siding Spring Observatory, Research School of Physical Sciences. Australian National Universities.

GIACCONI, R., H. GURSKY, E. KELLOGG, R. LEVINSON, E. SCHREIER & H. TANAN- BAUM. 1973. Astrophys. J. 184: I , 227.

PARKINSON, J. H. 1974. Nature (London) 249: 746. TARTER, C. B., W. H. TUCKER & E. E. SALPETER. 1969. Astrophys. J. 156:943.

D. Reidel Publishing Co. Dordrecht, The Netherlands. To be published.

J. 173: 213.

N.Y.

Brussels, Belgium.

Brighton, England.


21. 22.

23.

PRINGLE, J. E. 1973. Nature Phys. Sci. 243: 90. MCCRAY, R. 1974. X-rays in Space, International Conference on X-rays in Space.

LUCY, L. B. & P. M. SOLOMON. 1970. Astrophys. J. 159: 879. Vol. I: 547. University of Calgary. Calgary, Alberta, Canada.

24. SCHREIER, E., R. LEVINSON, H. GURSKY, E. KELLOGG, H. TANANBAUM & R. GIAC- CONI. 1972. Astrophys. J. 172: L79.

Falmer, Brighton, England; Institute of Astronomy. Cambridge, England. 25. TSURUTA, S . & M. J. REES. 1975. Preprint. Astronomy Centre, University of Sussex.

26. GNEDIN, Y. N. & R. A. SUNYAEV. 1973. Astron. Astrophys. 25: 233. 27. DoxsEY, R., H. V. BRADT, A. LEVINE. G. T. MURTHY, S. RAPPAPORT & G. SPADA.

1973. Astrophys. J. Lett. 182: L25.

1974. Astrophys. J. Lett. 190: L109. 28. HOLT. S. S . , E. A. BOLDT, R. E. ROTHSCHILD, J. L. R. SABA & P. J . SERELEMITSOS.

Documents

HER X-1 AND CEN X-3 REVISITED