------------------------------------------------------------------------
From: "Hans A. Dr. Mayer-Hasselwander"  hrm@mpe.mpg.de 
To: gcnews@aoc.nrao.edu
Subject:"High-energy Gamma-Ray Emission from the GC"
MIME-Version: 1.0

\title{High-Energy Gamma-Ray Emission from the Galactic Center}

\author{H.A. Mayer-Hasselwander^1, D.L. Bertsch^2, B.L. Dingus^3, A. Eckart=
^1,=20
J.A. Esposito^{2,10},=20
R. Genzel^1, R.C. Hartman^2, S.D. Hunter^2, G. Kanbach61, D.A. Kniffen^4, Y=
.C.=20
Lin^5,
P.F. Michelson65, A. M=FCcke^1, C. von Montigny^6, R. Mukherjee^7, P.L.Nola=
n^5, M.=20
Pohl^8,
O. Reimer^1, E.J. Schneid^9, P. Sreekumar^{2,10}, D.J. Thompson^2
}

\institute{
^1 Max-Planck-Institut f=FCr extraterrestrische Physik, D-85748 Garching,=
=20
Germany
^2 NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA
^3 University of Utah, Salt Lake City, Utah, USA
^4 Dept. of Physics and Astronomy, Hampden-Sydney College, Hampden-Sydney,=
=20
VA 23943, USA
^5 W.W. Hansen Experimental Physics Laboratory and Department of=20
 Physics, Stanford University, Stanford CA 94305, USA=20
^6 Landessternwarte K=F6nigstuhl, D-69117 Heidelberg, Germany
^7 Barnard College & Columbia University, Columbia Astroph. Lab, New York,=
=20
NY 10027, USA
^8 Danish Space Research Institute, 2100 Copenhagen O, Denmark
^9 Northrop Grumman Corporation, Bethpage, NY 11714, USA
^10 USRA Research Associate=20
}

\begin{abstract}

The EGRET instrument on the Compton Gamma-Ray Observatory has observed the
Galactic Center (GC) region with good coverage at a number of epochs.  A st=
rong
excess of emission is observed, peaking at energies >500 MeV in an error ci=
rcle
of 0.2 degree radius including the position l=3D0 and b=3D0 .  The close co=
incidence
of this excess with the GC direction and the fact that it is the strongest
emission maximum within 15 degrees from the GC is taken as compelling evide=
nce
for the sources location in the GC region.  The history of the emission
intensity, observed over 5 years, leaves room for possible time variation;
however, it does not provide evidence.  The angular extent of the excess ap=
pears
only marginally compatible with the signature expected for a single compact
object.  The emission therefore may stem from one or more compact objects o=
r may
originate from diffuse interactions within 85 pc from the center of the Gal=
axy
at 8.5 kpc distance.  The spatial distribution of the emission does not
correlate with the details of the CO-line surveys.  Thus, in spite of the
existence of a strong emission peak, earlier conclusions based on an appare=
nt
gamma-ray deficit, postulating the masses of the wide- line clouds in the G=
C
area to be an order of magnitude lower than indicated by naive CO
interpretation, are supported.  However, the total gas mass in the Nuclear =
Bulge
(NB) derived from the gamma-ray emission is found to be in agreement with t=
he
mass which in recent studies has been derived from molecular-line and FIR
surveys.  The g-ray emission spectrum is peculiar and different from the
spectrum of the large-scale galactic diffuse emission.  A diffuse emission
scenario requires an enhanced and peculiar Cosmic Ray (CR) spectrum as sugg=
ested
for the electrons in the Radio Arc.  A compact sources model hints at an or=
igin
in pulsars.  While the spectrum suggests middle-aged pulsars like Vela, too=
 many
are required to produce the observed flux.  The only detected very young pu=
lsar,
the Crab pulsar, has an incompatible spectrum.  However, it is not proven t=
hat
the Crab spectrum is characteristic for all young pulsars:  thus, a single =
or a
few very young pulsars (at the GC not detectable in radio emission), provid=
ed
their gamma-ray emission is larger than that of the Crab pulsar by a factor=
 of
13, are likely candidates.  Alternatively, more exotic scenarios, related t=
o the
postulated central black hole or dark matter (neutralino) annihilation, may=
 be
invoked.
\end{abstract}












Key words: gamma rays - galactic center
Send offprint requests to:hrm@mpe.mpg.de

1 Introduction
Gamma-ray emission from the Galactic Disk in the 30 MeV to 30 GeV range has=
 been=20
observed by the=20
telescopes on the OSO-3, SAS-2, COS-B spacecrafts and lately by the EGRET=
=20
telescope on the=20
Compton Gamma-Ray Observatory (CGRO). The comparison of the observed emissi=
on=20
with models for=20
interaction of galactic Cosmic Rays (CRs) with interstellar matter ( Strong=
=20
1995; Hunter et al. 1997)=20
has confirmed the expectation that a large fraction of the Galaxy=92s gamma=
-ray=20
emission stems from=20
large-scale diffuse collision processes. The emissivity at a particular loc=
ation=20
usually is proportional to=20
the product of target-particle (atoms, photons) density and CR (nucleons,=
=20
electrons) density. The=20
distribution of the emission as recorded by the observing instruments is=20
determined by its actual spatial=20
distribution within the Galaxy (intensity, direction, distance) and also by=
 the=20
angular resolution of the=20
observing telescopes. An emission excess from the Galactic Center (GC) at 8=
.5=20
kpc distance needs to be=20
resolved within the intense and narrow band of foreground- and background=
=20
emission from the large-
scale Galactic Disk. With current or previous instruments it only will be=
=20
detectable if either the target-
particle density or the CR density or both are peaking sharply within a few=
=20
hundred pc from the GC.=20
The galactic CO-line survey (Dame et al. 1987), which throughout the Galaxy=
 had=20
been found to be a=20
good tracer for interstellar molecular gas, shows a strong and structured e=
xcess=20
originating from within=20
the Nuclear Bulge (NB) which extends over the innermost 300 pc in radius.=
=20
Postulating the CR density=20
within the NB to be not significantly lower than the average derived for a=
=20
galactocentric radius <5 kpc,=20
led to the expectation that a strong and structured gamma-ray-emission exce=
ss=20
should be observed,=20
originating from interaction of CRs with the massive molecular clouds, loca=
ted=20
within the NB as=20
indicated by the CO-line survey.
It was an astonishing result of the COS-B mission that no pronounced source=
=20
excess was observed in=20
the direction to the GC. It was shown by Blitz et al. (1985), that the abse=
nce=20
of a corresponding strong=20
gamma-ray peak implies that in the volume surrounding the GC within a few=
=20
hundred parsecs the H2 to=20
12CO ratio is most likely smaller by an order of magnitude than elsewhere i=
n the=20
Galaxy. This can be=20
understood when investigating the CO surveys in more detail: the CO-line=20
emission in the=20
neighbourhood of the GC shows high velocity and velocity dispersion, not re=
lated=20
to the galactic=20
rotation. This atypical emission is attributed to gas clouds in local orbit=
al=20
motion and high turbulence.=20
The clouds also are expected to be under extreme conditions like high tidal=
=20
forces and temperatures.=20
This causes the highly velocity-dispersed CO emission to be observed optica=
lly=20
much thinner, making=20
the usual conversion factor to H2 irrelevant (Stacy et al. 1987; Stark et a=
l.=20
1991). The presumably=20
increased metallicity in the relevant volume also suggests a much lower H2 =
to=20
12CO ratio. Therefore the=20
gas tracing molecular-line surveys, when naively interpreted, indicate far =
too=20
much gas and are of no=20
value for a gas mass estimate within 3 degrees longitude from the GC. Furth=
er,=20
at l=3D0  no reliable=20
information is obtainable for the amount of HI by 21 cm observations due to=
 the=20
lack of sufficient=20
galactic-rotation velocity and velocity-dispersion in the direction of the=
=20
line-of-sight. In addition, strong=20
self-absorption does affect the foreground- and background emission. Theref=
ore,=20
the amount of HI=20
possibly concentrated within 100 pc from the GC is relatively unknown. Rece=
nt=20
far-infrared (FIR)=20
observations by the Diffuse Infrared Background Experiment (DIRBE) on COBE=
=20
(Sodroski et al. 1994)=20
show a less pronounced and more narrow spike in the longitude profile throu=
gh=20
the GC compared to=20
CO. The comparison of the CO and FIR surveys therefore indicate that the 12=
CO to=20
dust ratio is=20
unusually high and also that the emission from dust appears to originate fr=
om a=20
smaller volume around=20
the GC. The indicated uncertainties in the gas to dust ratio and other=20
complications of the dust-tracing=20
FIR emission (temperature) also tend to overestimate the gas mass in the ce=
ntral=20
region.=20
Sodroski concluded that the gas mass is probably less than required to be=
=20
observable in gamma-rays by=20
COS-B, which according to Blitz et al. (1985) gave a surprisingly low upper=
=20
limit of 4x10-7 photons=20
cm-2 s-1 for a point-source at the GC and energies >300 MeV. We have=20
reinvestigated the COS-B data=20
by comparing the >300 MeV intensity profile given in Mayer-Hasselwander et =
al.=20
(1982) with the=20
corresponding EGRET intensity profile and find fully compatible excesses fo=
r=20
both profiles within 3=20
degrees longitude from the GC. We can not reconstruct the COS-B result from=
=20
Blitz et al. (1985): it=20
appears too low by at least a factor 2.5. According to our analysis of COS-=
B and=20
EGRET data, a=20
correspondingly higher GC source flux on top of the predicted=20
large-scale-galactic diffuse emission=20
(Hunter et al. 1997) can be accommodated and thus the observed gamma-ray=20
emission constrains the=20
amount of diffuse emission from the GC less than do the FIR observations.
As previous COS-B analyses did not indicate a =91localised=92 source excess=
 and gas=20
mass estimates had=20
been reduced also on other arguments, it again was an unexpected result tha=
t=20
EGRET found a=20
pronounced source excess at the GC position (Mayer-Hasselwander et al. 1992=
,=20
1993) which is=20
designated in the second EGRET catalog (Thompson et al. 1995) by 2EG J1746-=
2852.=20
The EGRET=20
detection of a source at the GC is based on energies above 100 MeV; the loc=
ation=20
of the source is best=20
determined in the energy ranges above 500 MeV and there is found perfectly=
=20
compatible with the GC=20
(Lamb & Macomb 1997, Reimer et al. 1997). The high-energy gamma-ray measure=
ments=20
by EGRET=20
are of great importance for the understanding of the innermost part of our=
=20
Galaxy at several scales=20
within 1 kpc. It is evident that narrowing down the range of possible scena=
rios=20
by demonstrating either=20
stability or variability of the emission, by confining the extent and locat=
ion=20
of the source and by=20
determining the source spectrum in detail is essential for the scientific v=
alue=20
of this unexpected EGRET=20
discovery.
It is the topic of this paper to present the observational details and to d=
erive=20
the constraints which the=20
EGRET observations put on the nature of the source excess in the GC. The GC=
=20
source had to be=20
considered as possibly extended and confused and embedded in high and struc=
tured=20
background; this=20
required the application of analysis procedures partially different from th=
ose=20
used in EGRET standard=20
analyses.

2 Data and Analysis
2.1 EGRET Data
EGRET gamma-ray data relevant for the analysis of the GC region from observ=
ation=20
cycles 1, 2, 3, 4=20
and part of 5 are used. Table 1 lists the observation periods and their=20
parameters. For the purpose of=20
time variation analysis short observations at neighbouring epochs were grou=
ped=20
into clusters and were=20
combined in order to get statistically significant results. For the purpose=
 of=20
structural and spectral=20
investigations the data from all observations were added.
The analysis started with the standard EGRET calibration and exposures file=
s.=20
The latter contain=20
sensitivity corrections for each viewing period. These corrections to expos=
ures=20
for individual=20
observation periods in each energy range were derived (Esposito et al. 1998=
) in=20
order to account for in-
flight variation of sensitivity, primarily caused by the intrinsic ageing o=
f the=20
spark-chamber gas. In this=20
procedure, regions dominated by diffuse (and therefore non-variable) emissi=
on,=20
that were observed in=20
more than one observation at different epochs, were compared. The standard=
=20
corrections were derived to=20
normalise the average intensity for the whole field of view of an observati=
on=20
period to observations=20
when the sparkchamber performance was close to its originally-calibrated le=
vel.=20
This normalisation=20
postulates smooth variation versus epoch, i.e. no significant changes in=20
performance on short timescales,=20
except when a gas change took place. On a large scale, this correction proc=
edure=20
provides a good fit to=20
the instrument performance.
For a particular direction in the sky, some additional irregular sensitivit=
y=20
variations are seen. Because=20
the GC is a particularly complicated and interesting region, additional=20
constraints and corrections for in-
flight effects beyond the standard corrections were applied:=20
1. Only gamma rays incident within 25 degrees from the instrument pointing=
=20
direction were accepted, in=20
order to minimise source to instrument-axis offset-angle effects. The stand=
ard=20
EGRET analysis extends=20
to 30 degrees.=20
2. For each GC observation and for each energy range, additional correction=
=20
factors were derived on the=20
basis of the non-variable, intense galactic diffuse emission observed from=
=20
directions less than 20 degree=20
from the GC direction. Essentially, the data from each observation period w=
ere=20
normalised in such a=20
way that the normalisation is optimum for the GC location. For energies gre=
ater=20
than 100 MeV these=20
additional corrections in most cases are less than 15 percent. At lower ene=
rgies=20
in a very few cases in=20
phase 4 and 5 data do these corrections reach a factor of 2; however, these=
=20
observations have relatively=20
low exposures and correspondingly low statistical weights in the analysis a=
nd=20
thus their uncertainties do=20
not affect the results.
2.2 Diffuse Background Model
The galactic-diffuse-emission model developed by Bertsch et al. (1993) and=
=20
Hunter et al. (1994, 1995,=20
1997) is used to account for the large-scale galactic diffuse emission with=
in=20
which the GC source is=20
embedded.
This model is based on 12CO-line and HI surveys, a gas-coupled CR distribut=
ion=20
model, and an inverse-
Compton-emission model. Within 1 deg from l=3D0 , where the HI tracer measu=
rements=20
are incorrect due=20
to strong self-absorption and optical thickness effects (no velocity=20
dispersion), the HI data were=20
interpolated. In the range 354 to 10 degrees longitude the model disregards=
=20
untypical high-velocity 12CO=20
emission which is observed from the NB cloud complexes. Low velocity emissi=
on,=20
tracing the intercloud=20
medium and also low velocity emission from the clouds is entering the model=
 as=20
observed. Therefore, the=20
cloud complexes within 300 pc from the GC are not explicitly represented by=
 the=20
model. If emitting at a=20
detectable level, they are expected to be visible as gamma-ray excesses=20
exceeding the diffuse model.=20
Depending on their location, several clouds would be visible as individuall=
y=20
resolved sources and others=20
would contribute to an extended source excess.
The EGRET data are well represented by the diffuse emission model plus an=
=20
isotropic component=20
(Sreekumar et al. 1997) on a galactic scale. At specific locations, however=
,=20
deviations are observed.=20
Therefore, in individual smaller regions the representation can be improved=
 for=20
the purpose of=20
background estimation by additional local tuning in amplitude and bias. In=
=20
standard EGRET analysis=20
this tuning is implicit to the likelihood analysis algorithm =91LIKE=92 (Ma=
ttox et=20
al. 1996). In this analysis=20
the tuning was done as follows:
1.  The model was tuned in each energy range to fit the gamma-ray data best=
 in=20
the region within 20=20
degrees from the GC. The constraint placed on the tuning procedure was that=
 the=20
data could not be=20
exceeded by the model, such that the model forms a =91lower envelope=92,=20
representing the diffuse=20
emission, on which the emission excesses from individual sources are=20
superimposed.
2.  The galactic diffuse spectrum is known to be smooth. Thus, the tuning o=
f the=20
model has been iterated=20
for each energy range, aiming to make the spectrum of the background model=
=20
smooth (see figure 4).=20
This procedure minimised the statistical scatter of the fit for each energy=
=20
range.
Table 2 lists the tuning parameters applied to the EGRET diffuse model. The=
=20
intense diffuse emission=20
surrounding the GC region was therefore used as a fixed reference for both =
the=20
normalisation (section=20
2.1) and the background against which the GC region is viewed. This procedu=
re=20
removed much of the=20
ambiguity that otherwise affects analysis in this complicated region. The t=
uned=20
diffuse model was then=20
used as a constant background in the timevariation- and spectral- analysis =
and=20
as the reference for=20
comparison of the derived GC source-spectrum with the spectrum of the diffu=
se=20
emission.
2.3 Analysis
The analysis of a tentative source at the GC is difficult because a very di=
stant=20
galactic location is=20
observed within the high background due to the galactic diffuse emission=20
accumulating along the long=20
line of sight in front of and beyond the GC.
Due to the limited angular resolution of the instrument, source confusion i=
s a=20
concern particularly at=20
lower energies where the point spread function (PSF) extends over several=
=20
degrees. The identification of=20
the GC source as a single compact object therefore is possible only if=20
variability is observed. If a=20
spectrum distinctly different from that of the diffuse background emission =
is=20
observed, then at least it=20
can be demonstrated that the emission originates from a collection of pecul=
iar=20
sources or in an unusual=20
diffuse emission scenario. We address these topics, along with the source=
=20
significance, its possible=20
extent, and its location, using the following methods:
1.  The source location and significance is determined in four broad energy=
=20
bands, 30-100, 100-300,=20
300-1000 and >1000 MeV, accounting for the modelled diffuse background and =
using=20
likelihood=20
techniques.
2.  The extent of the GC excess has been compared, on the basis of maps wit=
h 0.1=20
degree binning, with=20
the extent of the source excesses observed for the three strongest pulsars:=
=20
Vela, Geminga and Crab,=20
which define the in-flight PSF. This analysis uses only the data above 1 Ge=
V=20
where the PSF is=20
sharpest.
3.  For the analysis of possible time variation, many observation periods=
=20
distributed over several years=20
had to be analysed individually. To obtain sufficient statistics for=20
investigation of time-variability,=20
three relatively wide energy bands were used: 100-300, 300-1000 and >1000 M=
eV.=20
For the time-
variation analysis a forth energy band (30-100 MeV) is rather useless due t=
o=20
insufficient angular=20
resolution with respect to the given background situation.
4.  For the determination of the source spectrum, the counts from the GC so=
urce=20
region have been=20
analysed in 10 smaller energy bins from 30 MeV to 10 GeV by evaluating maps=
 and=20
profiles of=20
counts and intensity directly as well as by using crosscorrelation- and=20
likelihood- analysis (Hermsen=20
1980, Mattox et al. 1996) methods. Cross-calibrations between the different=
=20
methods, especially=20
with respect to in-flight uncertainties in the PSF, have been performed aga=
in=20
for each energy range=20
by comparison with the analysis of the 4 strongest pulsars (including PSR=
=20
1706-44). Finally the GC=20
source-flux values in the low energy ranges were derived by likelihood anal=
ysis;=20
in the energy ranges=20
above 150 MeV the flux values were derived by a specific method of directly=
=20
accumulating the=20
source intensity within a small region (2 to 4 degree squares) of the binne=
d=20
data around the source,=20
together with applying an energy dependent normalisation factor. This appro=
ach=20
was found to be less=20
sensitive to the problems of high, not-perfectly-modelled background. The e=
rrors=20
are based on the=20
values obtained in a likelihood analysis and were scaled in accordance to t=
he=20
ratio between the source=20
counts obtained in this specific analysis and in the likelihood analysis.
The method applied was the only available method allowing for source confus=
ion=20
and a finite source=20
extent. It is to be noted, that the presence of a finite source extent or o=
f an=20
extended cluster of point=20
sources can make the available EGRET standard point-source analysis algorit=
hms=20
(using likelihood or=20
cross-correlation) inadequate because they are strictly based on the use of=
 the=20
PSF for counts=20
representing a single point source only.
Source confusion is a severe limitation in the interpretation of the result=
s.=20
The space volume which=20
might contribute unresolved sources to the observed excess increases drasti=
cally=20
at lower energies due to=20
the increasing size of the PSF. Thus the spectrum of a source at the GC loc=
ation=20
might be contaminated=20
severely by additional soft sources located within degrees from the GC at=
=20
energies below about 500=20
MeV. In this sense the observed spectrum of the GC source at energies below=
=20
about 500 MeV=20
represents an upper limit.

3 Results
3.1 Observational Scenario and Source Significance
There are 5488 =B1 516 counts observed for all energies above 30 MeV, repre=
senting=20
the total source=20
excess as seen by the analysis process described under 2.3. Illustrating th=
e=20
scenario, figure 1 displays=20
residual counts maps, after subtraction of the model-predicted diffuse emis=
sion,=20
in four energy ranges.=20
Statistical fluctuations are only slightly smoothed, retaining all possibly=
=20
significant structures in the=20
maps. In addition to the evident GC excess two prominent sources, the pulsa=
r PSR=20
B1706-44 near=20
l=3D343 , b=3D-3  and the quasar PKS 1622-297 near l=3D349 , b=3D13 , and s=
everal weaker=20
sources are=20
visible in the map. The number of counts observed for the sources are influ=
enced=20
by the exposure, which=20
varies through the map. However, the relative fluxes in the four energy ran=
ges=20
seen for each of these=20
sources already illustrate the fact that the GC excess and PSR B1706-44 hav=
e a=20
much harder spectrum=20
than the other sources.
3.2 Source Extent
The extent of the GC excess has been compared, on the basis of maps with 0.=
1=20
degree binning and=20
energies above 1 GeV where the angular resolution is best, with the extent =
of=20
the source excesses=20
observed for the pulsars: Vela, Geminga and Crab, which represent the in-fl=
ight=20
PSFs. While the=20
HWHM for the pulsars is 0.55 degree, the GC source excess has a HWHM of 0.7=
=20
degree. No=20
established method is available to quantify the properties of a possibly=20
extended source. We therefore=20
estimate, using the in-flight PSF and gaussian function approximations, for=
 the=20
GC source a spatial=20
intrinsic HWHM of 0.43 degree and a 68% flux enclosure angle of 0.6 degree.=
=20
Thus, the bulk of=20
emission from the GC source above 1 GeV is best compatible with emission=20
originating at 8.5 kpc=20
distance in a volume with a radius of 65 pc HWHM, respectively a volume wit=
h 85=20
pc radius for 68%=20
containment. However, in view of uncertainties due to  the high underlying=
=20
intense and structured=20
background, the source excess is still considered to be marginally compatib=
le=20
with emission from a=20
single compact object. A two-component model (see figure 5), where the emis=
sion=20
from a compact=20
source region (~60% from < 85 pc radius) is superimposed onto a resolved=20
extended emission excess=20
(~40% from < 250 pc radius) is representing the data even better. Further, =
an=20
extended source with 250=20
pc radius, for which the emissivity increases appropriately towards the cen=
ter=20
would have an=20
indistiguishable appearance.
3.3 Source Position
The positional analysis of the excess has been performed by cross-correlati=
on=20
and likelihood techniques,=20
assuming a single point source. The position-error contours are given in fi=
gure=20
2 for several energy=20
intervals above 100 MeV. Because there is evidence for extended emission fr=
om=20
the source, the=20
likelihood confinement contours are obtained too narrow, thus the discrepan=
cy=20
between the high and the=20
low energy ranges becomes less significant. Still, the shift in the maximum=
=20
supports the hypothesis that=20
the emission below 300 MeV includes flux from one or several other objects=
=20
located within 2 degrees=20
from the GC. At these lower energies an unidentified soft source in the sec=
ond=20
EGRET catalog, 2EG=20
J1747-3039 (Thompson et al. 1995), more recently designated as 3EG J1744-30=
11=20
with l=3D358.9, b=3D-
0.5 (Hartmann et al. 1998) may contribute. Several other objects in this re=
gion=20
(GRO 1744-28, 2S1743-
2941, E1740-2942, PSR1742-30, GRS1736-297, GRS 1739-278, GX359+02) might be=
=20
considered as=20
candidate counterparts, however, their spectra in energy ranges below gamma=
 rays=20
do not suggest their=20
visibility in the high-energy gamma-ray regime.
The emission center at energies above 1 GeV is confined to the GC direction=
 with=20
a < 0.2 degree radius=20
error box. This suggests that, at least at high energies, one source or ext=
ended=20
emission volume with a=20
very hard spectrum dominates the emission and is located or peaking within =
30 pc=20
from the actual GC.=20
In light of the absence of other strong gamma-ray sources within 15 degrees=
 of=20
this direction, the=20
statistical probability for this positional correlation to be a chance=20
coincidence is about 10-4, making it=20
very likely that the source is located near the GC and not anywhere else al=
ong=20
the line of sight.
3.4 Temporal Characteristics
The 25 observation periods, some of which had only low exposure, were combi=
ned=20
for time-variation=20
analysis into 10 datasets (designated as A -K in table 1 and figure 3). To=
=20
minimise statistical=20
uncertainties and at the same time keep spectral information and make use o=
f=20
good angular resolution at=20
high energies, three energy ranges were used:100 - 300 MeV, 300 MeV - 1 GeV=
 and=20
> 1 GeV. The time=20
history for the 3 energy ranges is displayed in figure 3. In addition to th=
e=20
flux values, also the applied=20
sensitivity corrections, the (accordingly corrected) exposures and the obse=
rving=20
angles of the source=20
relative to the instrument axis are displayed. It is conspicuous that the p=
oints=20
with the largest deviations=20
from the linear regression line tend to have larger error bars and also the=
=20
observational parameters are=20
less favourable. This correlation suggests that there are residual uncorrec=
ted=20
systematic effects. Even=20
with no fine-tuning Mc Laughlin et al. (1996) found only weak evidence for=
=20
variability of the GC=20
source. With the present more careful treatment of this specific region, th=
e=20
indication for variability is=20
even weaker. No long-term trend is seen, however, the possibility that a sh=
ort=20
term variation is indicated=20
by observation B (or H), where the flux in all three energy ranges is above=
=20
average, can not be excluded.
A detailed reanalysis of the COS-B data, taken more than an decade earlier,=
 was=20
not feasible. However,=20
the COS-B intensity profile given in Mayer-Hasselwander et al. (1982) for >=
 300=20
MeV has been=20
compared with a directly corresponding EGRET profile and showed agreement f=
or=20
the intensity within 3=20
degrees longitude from the GC, when normalised on the disk intensities obse=
rved=20
at 350 and 10 degree=20
longitude. We conclude that there is no evident difference in the intensiti=
es=20
which are observed on=20
average during both missions. The lack of detection of a GC source by COS-B=
=20
might be explained by=20
differences in the background model used, together with COS-B=92s much lowe=
r=20
sensitivity.
3.5 Energy Spectrum
The emphasis in evaluation of the spectrum has been given to the question: =
is=20
the spectral shape=20
different from that expected from interactions of ambient CRs with the=20
interstellar medium. The=20
spectrum of the diffuse emission is derived from the diffuse model (section=
 2.2)=20
for the region within 20=20
degree in longitude and latitude from the GC. The spectra for the GC source=
 and=20
for the diffuse=20
emission are compared in figure 4. They are seen to be distinctly different=
. It=20
is interesting to note that,=20
if neighbouring sources are involved at lower energies, being unresolved du=
e to=20
the broad PSF, then the=20
true spectrum of the high energy source excess at l=3D0 , b=3D0  will be ha=
rder and=20
even more different=20
from the diffuse emission spectrum. Previous analyses of the GC source spec=
trum=20
by Fierro (1995) and=20
Merck et al. (1996) were performed using EGRET standard analysis tools,=20
presuming an unresolved=20
point source. Those analyses already obtained rather hard spectra, somewhat=
=20
different from the diffuse=20
emission, however with a less pronounced peak and turnover at high energies=
.=20
This much more detailed=20
analysis clearly demonstrates that the spectrum of the GC source excess is =
not=20
due to ambient CR - gas=20
interaction. A very peculiar and very hard CR spectrum is required in order=
 to=20
explain the emission as=20
diffuse processes.
Allowing for a total source-excess extent up to 1.5 degree in radius, the f=
lux=20
attributed to the source=20
excess at l=3D0 , b=3D0  is (217=B115)*10-8 ph cm-2 s-1 for >100 MeV corres=
ponding to=20
a luminosity of=20
(2.2=B10.2)*1037 erg s-1 for a source distance of 8.5 kpc. =20
The fluxes in the energy bands are (in units of 10-8 ph cm-2 s-1): (95=B19)=
 (100 -=20
300 MeV), =20
(73=B14) (300 -1000 MeV), (49=B13) (> 1 GeV). The photon spectrum can be we=
ll=20
represented by a broken=20
power law with a break energy at 1900 MeV. =20
Below this energy the differential photon spectrum is=20
F(E)=3D(2.2=B10.01)*10-10(E/1900 MeV)-1.30=B10.03,=20
above the break energy the spectrum is F(E)=3D(2.2=B10.01)*10-10(E/1900=20
MeV)-3.1=B10.2.=20
The fluxes and errors do not reflect systematic uncertainties concerning=20
possible source confusion and=20
limitations of the background model and background subtraction method. The=
=20
likelihood of confusion=20
with possible unresolved soft-spectrum sources in the neighbourhood of the =
GC=20
increases to low=20
energies: the observed spectrum, when allocated to the high-energy source a=
t the=20
GC position therefore=20
must be considered to represent upper limits below ~300 MeV. In such a case=
 an=20
even harder spectrum=20
is required for the central source.
The integral flux for the GC source (exceeding the modeled diffuse emission=
)  in=20
the energy range > 100=20
MeV in this analysis is obtained larger by a factor 2 than the value given =
for a=20
point-source at the GC in=20
the second EGRET catalog. The difference is due to the different analysis=
=20
concepts: Here allowance is=20
made for an extended source and the data are analysed in narrow energy rang=
es,=20
not relying on details of=20
the PSFs; for the catalog, for which a standardised procedure had to be use=
d, a=20
point-source has been=20
postulated and a single wide energy range has been analysed by the likeliho=
od=20
procedure, trusting in the=20
PSF. Thus, the catalog flux does not contain the extended fraction of the e=
xcess=20
(see figure 5) and=20
furthermore, part of the flux is attributed to a neighbouring unidentified =
soft=20
source (see 3.3).

4 Source Scenarios
4.1 Interaction of Ambient Cosmic-Rays with Gas
In the GC (l=3D0 , Sgr A), as near l=3D0.6  (Sgr B), l=3D1.2  (Sgr D) and l=
=3D3  (Clump=20
2), there are situated=20
giant molecular cloud complexes. The masses of these complexes are not easi=
ly=20
determined and=20
originally had been very much overestimated on the basis of their signal in=
=20
12CO-line emission. Blitz et=20
al. (1985) and Stacy et al. (1987) on the basis of the non-detection of a=
=20
corresponding GC excess in=20
COS-B gamma-ray data have discussed the discrepancy between the 12CO (Dame =
et=20
al. 1987) tracer=20
prediction and the observed gamma-ray emission from these massive molecular=
=20
clouds. They already=20
suggested that the 12CO to H2 conversion factor for these clouds is very mu=
ch=20
lower than galactic=20
average. These cloud complexes are apparently in an unusual state of high l=
ocal=20
spherical and turbulent=20
gas velocities (making the 12CO line wide and optically thin), high molecul=
ar=20
gas kinetic temperature=20
and increased metallicity. Several authors (Nishimura 1980, Heiligman 1987,=
=20
Stark et al 1989, Cox &=20
Laureijs 1989, Pajot et al. 1989, G=FCsten 1989, Lis & Carlstrom 1994, Sodr=
oski et=20
al. 1994, Nakayama=20
et al. 1995) further investigated these clouds on the basis of tracer=20
molecule-line (13CO, CS, C18O) and=20
FIR- (IRAS, COBE-DIRBE, 900=B5) surveys and provide support for reduced mas=
s=20
estimates. Sodroski=20
et al. (1995) conclude that these complexes have 3 to 10 times less mass th=
an=20
originally suggested by=20
the 12CO emission and thus do not dominate the total gas mass in the centra=
l 300=20
pc. The possibility for=20
an even lower gas mass is indicated by CO-line observations of other=20
galaxy-cores by Wall et al.=20
(1989), who find CO/H2 conversion factors 5 to 20 times lower than our Gala=
xys=20
average. Stark et al.=20
(1989) suggested that the mass in the NB to a large extent consists of a di=
ffuse=20
molecular inter-cloud=20
medium, only partially bound to clouds.=20
Even when based on FIR observations, the average diffuse molecular gas surf=
ace=20
densities in the NB are=20
still high (=BB100 H-atoms cm-3, G=FCsten 1989). However, the NB region is =
much more=20
evolved compared=20
to the inner Galaxy, the advanced state of astration being reflected by the=
=20
chemical composition=20
(Wannier 1989). Thus, like the CO to gas, the FIR-emission to dust and the =
dust=20
to gas conversion=20
factors are subject to uncertainties in this region and are possibly lower =
than=20
those used in the gas=20
estimate by Cox & Laureijs (1989). It is noted that that the longitude=20
distribution of FIR emission=20
IRAS, COBE-DIRBE) within the NB, in contrast to CO-line distributions, is p=
eaked=20
toward the=20
longitude l=3D0  and appears only slightly more extended in longitude than =
the=20
gamma-ray distribution.=20
However, the absolute amount of gas (R<250 pc) again remains uncertain due =
to=20
the difficulties in=20
determining the FIR to gas conversion factor in this particular  region.
It is emphasized that the gamma-ray data do not show any localized source=
=20
excesses at the locations of=20
the cloud complexes Sgr B, SgrC, Sgr D and Clump 2. This finding does not d=
epend=20
on the details of=20
the diffuse model which disregards most of the 12CO emission from these clo=
uds=20
(Bertsch et al., 1993).=20
The non-detection of these complexes also excludes the possibility of a=20
significantly enhanced CR=20
density within these clouds. Further it strongly suggests that the smaller=
=20
complexes Sgr A and Sgr C (if=20
there is not an enhanced CR density) also do not contribute significantly t=
o the=20
central source excess.
Figure 5 displays the enclosed-gas-mass estimate based on FIR observations =
by=20
Cox & Laureijs (1989)=20
depending on galactic radius (dotted line). The corresponding predicted=20
gamma-ray emission > 300=20
MeV, based on the average emissivity (1.05 10-26 atom-1 sr-1 s-1) derived f=
or=20
the inner Galaxy < 5 kpc by=20
Strong & Mattox (1995), is indicated by the left-hand scale. Also shown is =
the=20
EGRET-observed=20
gamma-ray emission: the circle symbol indicates the total flux observed fro=
m a=20
volume of  250 pc=20
radius around the GC. For the comparison of the observed gamma-ray emission=
 with=20
the emission=20
expected from the diffuse matter distribution, based on FIR, we use a=20
two-component model (see 3.2)=20
where an unresolved component (~60%, double-hashed area) originates from wi=
thin=20
85 pc and an=20
evidently resolved extended component (~40%, hashed area) is emitted from w=
ithin=20
250 pc from the=20
GC. The latter component partially is represented in the diffuse model and=
=20
partially is observed as an=20
excess exceeding the diffuse model. The observed unresolved gamma-ray sourc=
e=20
with its peculiar=20
spectrum can well be accommodated when a slightly reduced emissivity (~11 %=
) in=20
the NB relative to=20
the estimate from FIR (Cox & Laureijs,1989) is adopted. This minor differen=
ce=20
can as well be=20
attributed to the uncertainties in deriving the gas estimate from FIR as to=
 the=20
uncertainty in the=20
determination of the amount of gamma-ray emission already represented by th=
e=20
diffuse model or to a=20
lower emissivity within this region. Dahmen et al. (1997), on the basis of =
a=20
detailed analysis of their=20
C18O survey and from 12CO data obtain a somewhat lower value of  ~27 106 so=
lar=20
masses for the=20
volume within 300 pc from the GC.
4.2 Interaction of Enhanced Cosmic Rays with Gas or Interstellar Radiation
The peculiar spectrum of the of the observed gamma-ray emission (figure 4),=
 in=20
case of origin from CR=20
- gas interactions, requires a scenario where the CRs, nucleons or electron=
s,=20
have a very hard and=20
unusual spectrum and where the gamma-ray creation process basically preserv=
es=20
this spectrum. It is=20
difficult to produce the observed hard spectrum from nucleonic interactions=
, at=20
least for isotropic=20
interaction with the target medium where the =91po bump=92 is always produc=
ed.=20
Nothing is known about the=20
nucleonic CR density in the Galactic Bulge, and gamma-rays would be the onl=
y=20
tracers. If the gas=20
density increases toward the center and some coupling of CR density to the =
gas=20
density exists, then the=20
creation of the observed amount of gamma-rays might be feasible; however, t=
he=20
problem of producing=20
the observed peculiar spectrum remains.
Energetic electrons with hard spectra, and subsequent gamma-ray spectra, ca=
n=20
more easily be produced.=20
In contrast to CR nucleons, high energy electrons are well traced by their=
=20
non-thermal radio=20
(synchrotron) emission and are indeed observed particularly from the compac=
t=20
source Sgr A* and from=20
the extended radio Arc. Energetic electrons not only can interact via=20
Bremsstrahlung with nucleonic=20
targets, but also can create gamma-rays by inverse-Compton interaction with=
=20
radiation fields.
The Arc
An appropriate scenario is offered by the radio Arc, where the presence of=
=20
rather monoenergetic=20
electrons in an extended region is indicated by the observed radio spectrum=
=20
(Yusef-Zadeh 1989; Lesch=20
& Reich 1992; Pohl et al. 1992; Yusef-Zadeh & Wardle 1993). Pohl (1997) sug=
gests=20
that the electrons,=20
accelerated and confined in the extended flux tube of the Arc, interact wit=
h the=20
FIR radiation field=20
originating in a neighbouring massive cloud (M0.20-0.033). This cloud is in=
=20
contact with the flux tube=20
through an HII region (G0.18-0.04, the =91Sickle=92), which also could act =
as the=20
source injecting electrons=20
into the flux tube. The most energetic of the accelerated electrons can=20
up-scatter FIR photons to gamma=20
rays. Pohl=92s model successfully produces the observed spectrum and intens=
ity.
Sgr A*
The spatial distribution for the interstellar radiation field (star light a=
nd=20
FIR reprocessed from dust) near=20
the GC has a HWHM of a fraction of a degree. It peaks sharply within the IR=
S16=20
cluster about 1 arcsec=20
west of Sgr A* (Menten et al. 1997), and therefore would provide an appropr=
iate=20
target distribution for=20
the creation of the observed gamma radiation. However, the quasi monoenerge=
tic=20
non-thermal electron=20
spectrum observed from Sgr A* (Duschl & Lesch 1994; Zylka et al. 1995; Beck=
ert=20
et al. 1996; Metzger=20
et al. 1996) is modelled to have a mean energy below 150 MeV and thus is=20
inadequate to produce the=20
observed gamma-ray spectrum.=20
4.3 Stellar Sources
Pulsars
A stellar source or a collection of stellar sources at a distance of 8.5 kp=
c=20
would need to have a=20
luminosity (> 100 MeV) about equivalent to that of 13 Crab pulsars, 50 Vela=
=20
Pulsars or 1170 Geminga=20
Pulsars to explain the unresolved fraction of the source excess. As a singl=
e=20
source, it would be by far=20
the most luminous source in the Galaxy. Radio emission from young pulsars c=
an=20
not be detected from=20
the GC region because the dispersion along the line of sight is too great, =
so=20
the non-detection in radio=20
does not exclude this possibility. From a spectral point of view, the older=
=20
(>>103 yr) pulsars like Vela or=20
Geminga, provide hard spectra with a break at a few GeV, resembling the obs=
erved=20
GC spectrum, while=20
the younger Crab pulsar has a spectrum which is too soft. However, the Crab=
=20
pulsar is the only example=20
available and one can argue that this single case does not exclude the=20
possibility that other very young=20
pulsars could have a spectrum similar to that of Vela. Thus, disregarding t=
he=20
spectral argument, one or=20
a few very young pulsars could be the sources of the observed emission, pro=
vided=20
their individual=20
gamma-ray-emissivity is higher by about an order of magnitude. This could b=
e=20
achieved by an increased=20
breaking energy loss combined with an increased efficiency by which the bre=
aking=20
energy is converted=20
into gamma-ray emission.
It is interesting to consider the alternative case, where the observed=20
luminosity may be produced through=20
an accumulation of many middle-aged pulsars which would provide the observe=
d=20
spectrum. Metzger et=20
al. 1996 conclude that star formation in the NB is not a continuous process=
, but=20
is rather episodic,=20
possibly recurrent, and that the last mild starburst occurred 107 yr ago.=
=20
Pulsars are suggested to have at=20
birth a kick-off velocity distribution with a mean velocity of 450 km s-1 (=
Lyne=20
& Lorimer 1994). A=20
fraction of the pulsars evolving from the starbursts in the GC region is ej=
ected=20
with favourable initial=20
velocity vectors (low speed, retrograde ejection) forcing them into low=20
Keplerian orbits around the GC=20
due to the central gravity potential well. This collection of pulsars possi=
bly=20
forms a local (< 1 pc)=20
concentration of sources around the GC which is unique within the Galaxy.=
=20
However, the number of=20
pulsars which have an age enabling them to emit significant gamma-ray flux=
=20
appears far too small to=20
make this scenario likely.
Massive Black Hole
Accretion to the postulated central massive black hole (BH) can not easily=
=20
produce the observed gamma=20
radiation. The likely BH of 2.45 106 solar masses (Eckart & Genzel 1996),=
=20
postulated to resemble the=20
dark mass located at the GC, which is possibly identical with Sgr A*, appar=
ently=20
is in a =91starving state=92.=20
Besides synchrotron radio/IR emission, not much radiation is observed. Sgr =
A* is=20
only a weak emitter in=20
soft X-rays (Goldwurm et al. 1994) and has never been detected above 30 keV=
 in 6=20
years of  SIGMA=20
data (Vargas et al. 1997). The weak X-ray emission makes it unlikely that t=
he=20
gamma-ray source is=20
related to the wind accretion from the IRS16 cluster to the tentative BH as=
=20
suggested by Melia (1992)=20
or Mastichiades & Ozernoy (1994). More recently the concept of=20
advection-dominated accretion-flow=20
(ADAF) into BHs has been suggested (Mahadevan et al. 1997). Markoff, Melia =
and=20
Sarcevic (1997)=20
performed a refined calculation of the particle cascade and explain the obs=
erved=20
spectrum to some=20
extent by a combination of proton-synchrotron radiation and pion decay. How=
ever,=20
the emission in the=20
high-energy gamma-ray regime in all models is dominated by pion -production=
 and=20
-decay and therefore=20
the models do not easily produce the very hard spectrum with its sharp turn=
over,=20
which is observed for=20
the GC source. Clearly emission in the vicinity of a supermassive BH remain=
s a=20
possibility; however its=20
feasibility still is not convincingly demonstrated.4.4 Other Scenarios
Dark Matter
The =91neutralino=92 species of the WIMPs (weak interacting massive particl=
es) are=20
the lightest particles=20
within the SUSY model of particle physics and are expected to concentrate i=
n the=20
centers of galaxies=20
forming non-dissipative gravitational singularities (NGS). These gravitatio=
nal=20
potential wells may act as=20
seeds for generating massive BHs. The sharp increase in neutralino density =
near=20
the singularity leads to=20
neutralino annihilation, which might be traced by decay products including =
gamma=20
rays in the MeV to=20
TeV range (Silk & Bloemen 1987, Stecker & Tylka 1989; Berezinsky et al. 199=
2;=20
Urban et al. 1992).=20
The spectra expected from annihilation of neutralinos, particularly of phot=
inos,=20
higgsinos, etc. in the GC=20
consist of a flat continuum in the range above 100 MeV, dropping exponentia=
lly=20
at several GeV, and=20
also containing a line corresponding to the neutralino mass; possible value=
s for=20
the neutralino mass=20
range from 5 GeV to several 100 GeV. Stecker & Tylka (1989) obtain for a=20
neutralino mass of 15 GeV=20
a spectrum which is compatible with the observed GC source spectrum. They u=
se=20
the isothermal model=20
for the dark-matter-core by Ipser and Sikivie (1987), which provides a cent=
ral=20
core density distribution=20
compatible with the observed source, and predict a gamma-ray intensity whic=
h is=20
a factor 4 less than=20
observed. This difference can be considered small with respect to the avail=
able=20
freedom in the choice of=20
model parameters. However, Berezinsky et al. 1992, on the basis of the radi=
o=20
emission expected from=20
electrons, which also emerge from the decay processes, argue for a lower=20
neutralino mass limit of >150=20
GeV yielding much lower intensities in the GeV range. So the possibility th=
at=20
the GC source represents=20
neutralino decay remains speculative; the safest indication that such proce=
sses=20
are actually occuring=20
within the NB could come from the detection of the accompanying neutralino=
=20
annihilation-line which is=20
expected within the energy range from 10 GeV to a few 100 GeV.

5 Summary
A strong high-energy gamma-ray excess, on top of the expected galactic diff=
use=20
emission is observed to=20
originate from a volume of typically 85 pc or 0.6 degree radius peaking wit=
hin=20
30 pc or 0.2 degrees=20
from the actual GC. The emission volume is likely extended or contains mult=
iple=20
sources; however, it is=20
still marginally compatible with a point source. The observation is well=20
described by an unresolved=20
emission peak, located on top of an extended (resolved) emission excess whi=
ch=20
representis about 40% of=20
the total flux emitted from the  volume with radius 250 pc around the GC; t=
he=20
latter extended=20
component could represent the diffuse emission expected from interaction of=
=20
interstellar gas and inner-
galactic average-density Cosmic Rays. The source flux appears to be stable =
over=20
the EGRET=20
observation epochs: some indication for short term variation within a facto=
r of=20
2 exists, but this=20
apparent variation is more likely due to systematic uncertainties in the=20
measurement. A reanalysis of the=20
earlier COS-B data showed that the intensity observed by COS-B from the GC =
is=20
compatible with that=20
observed by EGRET. The spectrum is very hard up to 2 GeV, where it turns ov=
er=20
sharply; thus it is=20
distinctly different from the spectrum of the large-scale galactic diffuse=
=20
emission, within which the GC=20
source is embedded.
The earlier finding that the non-detection by COS-B of a prominent gamma-ra=
y=20
source at the GC=20
requires lower masses for the molecular cloud complexes in the NB than indi=
cated=20
by naive=20
interpretation of other gas tracers is confirmed, in spite of the detection=
 of=20
the strong GC source excess=20
by EGRET. However, this confirmation is now based on the longitude distribu=
tion,=20
which does not trace=20
the cloud complexes seen in CO, and on the peculiar spectrum of the observe=
d=20
emission. The total gas=20
mass within 250 pc from the GC derived in this analysis is found to be=20
compatible with the estimates=20
from FIR surveys.
For the central unresolved source excess, based on the observed peculiar=20
spectrum, on the limited extent=20
of the source and on the apparent lack of time variation, three different s=
ource=20
scenarios are favoured=20
candidates:=20
  A single or a few very young pulsars (<1000 yr), which would need both a =
high=20
rate of energy loss=20
and a very high gamma-ray emission efficiency, could produce the observed s=
ource=20
excess,=20
equivalent to the emission of ~13 Crab pulsars. This scenario postulates th=
at=20
the observed spectrum=20
of the Crab pulsar is not representative for the spectral properties of ver=
y=20
young pulsars.
  Radio spectra indicate the presence of high-energy quasi-monoenergetic=20
electrons in the flux tube of=20
the radio Arc structure near the GC. These electrons interact with the FIR=
=20
radiation field of the=20
neighbouring massive molecular cloud and could emit gamma-rays by=20
inverse-Compton boosting of=20
the FIR photons, providing a compact gamma-ray source with the observed spe=
ctrum=20
and intensity.
  WIMPs are hypothesised to concentrate in the center of the Galaxy forming=
=20
dark-matter cores with=20
densities high enough to trigger neutralino annihilation. The annihilation=
=20
processes produce, through=20
several decay channels, gamma rays with a continuum spectrum above 100 MeV =
and=20
an annihilation=20
line corresponding to the neutralino mass. The spectrum and expected intens=
ity=20
depends strongly on=20
the neutralino mass, which is unknown. The source extent is expected to be=
=20
compatible with the=20
observation for a wide range of core-model parameters.
To proceed significantly further, new observations in the 100 MeV to 100 Ge=
V=20
regime, with higher=20
spatial resolution and higher sensitivity, are required aiming to determine=
=20
better the extent and possible=20
time variation of the source and the high-energy continuation of the spectr=
um.=20
This task will have to=20
await a next-generation high-energy instrument such as GLAST. Earlier progr=
ess=20
might arise from the=20
atmospheric Cerenkov telescopes sensitive below 1 TeV just becoming operati=
onal=20
or under=20
construction; however, the steep spectral decrease of the spectrum above 2 =
GeV=20
might inhibit the=20
detection.


Acknowledgements
The EGRET team gratefully acknowledges support from the following:=20
Bundesministerium f=FCr Bildung,=20
Wissenschaft, Forschung, und Technologie (BMBF), Grant 50 QV 9095 (MPE auth=
ors),=20
NASA Grant=20
NAG 5-1742 (HSC); NASA Grant NAG-1605 (SU); and NASA Contract NAS 5-31210 (=
GAC).

References

Beckert, T., et al. 1996, A&A 307, 450
Berezinsky, V.S., Gurevich, A.V. and Zybin, K.P. 1992, Phys.Lett. B294, 221
Bertsch, D.L., et al. 1993, ApJ 416, 587
Blitz, L., et. al. 1985, A&A 143, 267
Cox P. & Laureijs, R. 1989: The Center of the Galaxy, ed. M. Morris, Kluwer=
,=20
Dordrecht, IAU Symp.=20
136, pp. 121
Dahmen G., et al. 1997, A&AS 126, 197
Dame T.M., et al. 1987, ApJ 322, 706
Duschl, W.J. & Lesch, H. 1994, A&A 286, 431
Eckart, A., et al. 1993, ApJ 407, L77
Eckart, A. & Genzel, R. 1996, Nature 383, 415
Esposito, J.A. 1998, in preparation
Fierro, J.M. 1995: Observ. of Spin-Powered pulsars with EGRET, Thesis, Stan=
ford=20
University
Genzel, R. and Townes, C.H. 1987, ARA&A 25, 377
G=FCsten, R. 1989: The Center of the Galaxy, ed. M. Morris, Kluwer, Dordrec=
ht, IAU=20
Symp. 136, pp. 89
Goldwurm, S.A., et al. 1994, Nature 371, 589
Heiligman, G. 1987, ApJ 314, 747
Hartmann, R.C., et al. 1998, ApJS, in preparation
Hermsen, W. 1980, Gamma-Ray Sources, Thesis, Rijksuniversiteit Leiden
Hunter, S.D., et al. 1994, ApJ 436, 216
Hunter, S.D., et al. 1995: Proc. 24th ICRC 2, 182
Hunter, S.D., et al. 1997, ApJ, in press
Ipser, J.R. & Sikivie, P. 1987, Phys. Rev. D35, 3695
Krichbaum, T.P., et al. 1993, A&A 274, L37
Lamb, R.C. & Macomb,D.J. 1997, ApJ 448, 872
Lesch, H. & Reich, W. 1992, A&A 264, 493
Lis, D.C. & Carlstrom, J.E. 1994, ApJ 424, 189
Lyne, A.G. & Lorimer, D.R. 1994, Nat 369, 127
Mahadevan, R., Narayan, R., Krolik, J. 1997, ApJ 486, 268
Markoff, S., Melia, F., Sarcevic, I. 1997, ApJ 489, L47
Mastichiadis, A. & Ozernoy, L.M. 1994, ApJ 426, 599
Mayer-Hasselwander, H.A., et al. 1982, A&A 105, 164
Mayer-Hasselwander, H.A. 1992: Talk given at First Compton Symposium, St. L=
ouis
Mayer-Hasselwander, H.A. 1993: Proc. 23rd ICRC, Calgary, 1, pp. 47
Mattox, J.R. 1996, ApJ 461, 396
Melia, F. 1992, ApJ 387, L25
Menten, K.M., et al. 1997, ApJ 475, L111
Merck, M., et al. 1996, A&AS 120, 465
Mezger, P.G., et al. 1996, A&AR 7, 289
Nakagawa, T., et al. 1995, ApJ 455, L35
Nishimura, T., et al. 1980, ApJ 239, L101
Pajot, F., et al. 1989, A&A 223, 107
Pohl, M., et al. 1992, A&A 262, 441
Pohl, M. 1997, A&A 317, 441
Reimer, O., Dingus, B.L., Nolan, P.L. 1997, Proc. 25th ICRC, Durban, 3, pp =
97
Silk,J., Bloemen H. 1987, ApJ 313, L47
Sodroski, T.J., et al. 1994, ApJ 428, 638
Sodroski, T.J., et al. 1995, ApJ 452, 262
Sreekumar, P. 1997: Proc. 4th Compton Symposium, Williamsburg
Stark, A.A., et al. 1989, The Center of the Galaxy, ed. M. Morris, Kluwer,=
=20
Dordrecht, IAU Symp. 136,=20
pp. 129
Stacy, J.G., et al. 1987, Proc. 20th ICRC, 1, 17
Stark, A.A., et al. 1991, MNRAS 248, 4
Stecker, F.W. & Tylka, A.J. 1989, ApJ 343, 169
Strong, A.W. 1996, Space Science Rev. 76, 205
Strong, A.W. & Mattox, J.R. 1996, A&A 308, L21
Thompson, D.J., et al. 1995, ApJS 101, 259
Urban, M., et al. 1992, Phys. Lett. B 293, 149
Yusef-Zadeh, F. 1989, The Center of the Galaxy, ed. M. Morris, Kluwer,=20
Dordrecht, IAU Symp. 136,=20
pp. 243
Yusef-Zadeh, F. & Wardle, M. 1993, ApJ 405, 584
Vargas M., et al. 1997, ApJ 476, L23
Wall, W.F., et al. 1993, ApJ 414, 98
Wannier, P.G. 1989, The Center of the Galaxy, ed. M. Morris, Kluwer, Dordre=
cht,=20
IAU Symp. 136,=20
pp. 107
Zylka, R., et al. 1995, A&A 297, 93



Galactic Center paper

Figures and Tables

Table 1: Observation Period Parameters

Table 2: Diffuse Background Model Parameters

Figure 1: Residual smoothed counts maps and profiles in several energy rang=
es=20
after=20
subtraction of the model-predicted diffuse emission background. The panels =
refer=20
to these=20
energy ranges:=20
a) > 1 GeV, b) 300 MeV - 1 GeV, c) 100 - 300 MeV, d) 30 - 100 MeV.

Figure 2: The contour lines indicate the 50%, 68%, 95%, and 99% containment=
=20
probability=20
regions for the source position as derived by the EGRET likelihood analysis=
=20
program.

Figure 3: Time history of the observed flux at ten epochs. Panel a) display=
s the=20
exposure=20
correction factor which was applied to correct for in-flight sensitivity=20
changes. Panel b) gives=20
the corrected exposure. Panel c) shows the offset of the GC direction relat=
ive=20
to the instrument=20
axis. The panels d) display the observed source fluxes in three energy rang=
es=20
and=20
corresponding linear regression lines. The data point for observation K at =
100=20
to 300 MeV is=20
out of range, it is a factor 2 larger in value and error than data point H,=
 but=20
is considered=20
unreliable due to the short exposure and large offset angle.

Figure 4: The lower panel gives the differential photon spectrum for the GC=
=20
source excess=20
together with a broken-power-law fit. The upper panel shows the power (per =
ln(E)=20
interval)=20
spectrum for the source compared with the spectrum of the embedding galacti=
c=20
diffuse=20
emission averaged in the longitude range 340 - 20 degrees.

Figure 5: The spatial distribution of the observed gamma-ray source compone=
nts=20
are=20
compared with the predicted emission based on the gas estimate from FIR sur=
veys=20
(dotted line)=20
and on the average CR density within a galactocentric radius of 5 kpc. The=
=20
filled circle gives=20
the total observed flux from within 250 pc around the GC. The hashed areas=
=20
indicate the=20
components of the total flux: the fraction beeing already represented by th=
e=20
diffuse model, the=20
extended component


16




------------- End Forwarded Message -------------



------------- End Forwarded Message -------------