If data were available for individual detectors or quadrants, then three lightcurves were produced for each energy range and time interval; one for the two aligned quadrants and one for each of the offset quadrants (since they pointed in different directions on the sky). If data were only available for each half-array, then separate lightcurves for the aligned and offset quadrants were produced for each energy range. In this case, a source detected in an offset quadrant could not always be unambiguously located on the sky, since the counts may have arisen from one of two separate locations. In the cases when a single data stream covering the entire detector was selected, only a single set of lightcurves covering both the aligned and offset quadrants was produced. The time taken for a source to pass through a FOV is dependent on the slew rate, which was nominally 42hr-1, with two faster speeds of 85 and 170 hr-1 also used. At the slowest rate, a source would take 135s to pass completely through the FOV, from zero response to zero response. The 1-8 keV energy range was chosen since this is where the residual background in the ME was lowest and the sensitivity highest for a source with a typical X-ray spectrum. The 10-18 keV energy range was chosen to sample the solar particle induced background counting rate; the contribution of most X-ray sources in this energy range being small. The slew lightcurves were accumulated with a binning time of 10 s. This was chosen to allow a uniform set of lightcurves to be produced, since all the ME data configurations provided spectra with 10 s or faster time resolution, while allowing the collimator profile to be adequately sampled while slewing.
|Figure 3: Exposure map for the complete EXMS catalog in galactic coordinates. Areas which were never slewed over, or had exposures of <3 s are shaded black. All colored regions received an exposure of 3 s. Areas colored white received exposures >600 s. 98% of the sky was slewed over at least once and 85% of the sky received an exposure of >60 s|
A schematic diagram of a typical three-legged slew is shown in Fig. 4. The direction of slew is always parallel to one edge of the rectangular FOV. In Fig. 5 the region of the sky swept out by a real three-legged slew is shown. In the first, second, and third legs, the FOVs swept out two, three and two strips of sky, respectively. The differing paths swept out by the aligned and offset quadrants are shown, together with the positions of the eight detected sources.
|Figure 4: Schematic diagram of a three-legged slew, with the four quadrants of the ME arranged in the usual combination of two aligned and two offset. The squares indicate the FOVs of each quadrant and are not to scale. Slew directions were always parallel to one edge of the FOV|
|Figure 5: The "footprint'' of a representative three-legged slew showing the area of sky observed. The hatched and cross-hatched areas show the area covered by the offset and aligned quadrants, respectively. The start of the slew was in the lower left corner. The variation in the width of the footprint results from the projection onto equatorial coordinates. The positions of the sources identified in Fig. 6 are marked with stars|
To illustrate the quality and type of data available in the EXMS, Fig. 6 shows representative 1-8 keV lightcurves for a slew across a crowded region of sky (the same slew as illustrated in Fig. 5). The three panels show lightcurves for the aligned and two offset quadrants. One source (4U1627-673) is seen in the two offset quadrants at different times (separated by 240 s). All count rates are normalized to s-1 half-1 (4 detectors), but no correction for collimator losses has been applied. The lightcurves are not background subtracted and the nominal background counting rate of 15 s-1 half-1 is visible whenever there are no sources in the FOV. The 3 slew legs can be seen in the lightcurves since there are missing data between the slew legs (just after 16 hr and just before 16.6 hr). This is due to data from contiguous intervals of less than 5 min duration not being stored on EXOSAT data tapes.
The slew lightcurves were systematically searched for sources passing
through the FOV by modeling the observed time-dependent
1-8 keV count rate, R1-8, by:
The source term depends on the source count rate, S, the time of transit through the collimator center, , and the collimator profile determined by the slew rate. The 1-8 keV background count rate was assumed to be linearly related through a constant, k, to the 10-18 keV count rate, R10-18. This was verified by inspection of the lightcurves which showed that only sources with peak count rates 100 s-1 were detected in the 10-18 keV energy range. This approach is successful in excluding the type of background events illustrated in the upper panel of Fig. 6, in which short-term increases in the solar particle flux can mimic the intensity profile of an X-ray source.
For a fixed best-fit parameters S and k were determined by maximum likelihood. Candidate detections were initially identified in a coarse search in , after which parameters and their uncertainties were precisely calculated. The transit of some strong sources can be located to <0.1 s. Occasionally, it was necessary to fit two sources simultaneously. The limiting 1-8 keV sensitivity, when no confusing sources are present, is estimated to be erg cm-2 s-1 at the slowest slew rate. The source best-fit locations and error boxes were reconstructed using the values of and by parabolic interpolation from the positions contained in the pointing files. A total of 1210 detections were made in this way and are shown in galactic coordinates in Fig. 9.
If the fit to the collimator profile is formally acceptable, the time of maximum intensity corresponds to the position of the quadrant when the source passed through its midline. Hence, source positions can be well localized in the direction of slew, but the position perpendicular to the slew axis is constrained only by the width of the FOV, unless assumptions about the source intensity are made. This means that, for the majority of sources, the uncertainty regions are narrow in the slew direction and much broader perpendicular to it.
The mean length of the narrow side of the 1210 uncertainty regions is (6.30 0.12)'; (all errors are given at 68% confidence). In Fig. 7 a schematic diagram of a typical EXMS uncertainty region is shown defining the different offsets which are referred to later. In Fig. 8 the lengths of the narrow sides are plotted against uncorrected source counts for the 1210 detections. This demonstrates that the brightest detections tend to have the narrowest uncertainty regions, as expected. Since these regions are constructed under the assumption of a constant source intensity, it is to be expected that some identifications will lie outside their formal uncertainty region (see Fig. 7). Also seen in Fig. 8 is the effect of the different slew rates on the width of the uncertainty region for a given count rate, leading to three distinct clusterings of points, superimposed on the scatter caused mainly by source variability.
|Figure 6: Lightcurves obtained during a slew across a crowded region of sky, obtained from the two aligned quadrants (upper panel) and the two offset quadrants (middle and lower panels). No correction for collimator losses has been applied. Source identifications are given. The count enhancement at around 16.4 hr in the aligned quadrants is due to increased solar activity. The slew rates were 42hr-1, 85hr-1 and 42hr-1 during the first, second, and third slew legs, respectively. The intervals of missing data correspond to the breaks between slew legs. Note that 4U1627-673 was detected in both offset quadrants separated by 240 s|
|Figure 7: Schematic diagram of a formal EXMS uncertainty region, shown as a hatched rectangle within the rectangular FOV. The arrow shows the slew direction. The source positions I and E are internal and external to the uncertainty region, respectively. C is the uncertainty region centroid. EC and IC are the distances between sources E and I and the centroid. EM and IN are the parallel, or vector, offsets in the direction parallel to the slew. CM and CN are the distances which determine the correction for collimator transmission|
|Figure 8: Plot of the dimensions of the short sides of 1210 uncertainty regions against source count rate. The clustering of the data points into three distinct lines is a consequence of the three different slew rates, the slowest of which allowed sources to be localized most precisely|
|Figure 9: The 1210 EXMS detections in galactic coordinates. The predominance of galactic plane and Magellanic cloud sources is evident|
The sources found in the above searches are all within the formal uncertainty regions. However, apart from these "internal'' sources, attention was also paid to any bright, time variable and/or extended sources whose derived positions were outside the formal uncertainty regions, but which still passed within the FOV during a slew. Such sources, typically X-ray binaries or supernova remnants, may well have incorrectly determined uncertainty regions due to time-variability, or their extended nature. In practise, whenever this occured, there was little doubt that the detection was due to the "external'' source, since there was either no candidate in the uncertainty region, or a source not known to be a strong X-ray emitter.
Some sources are considered confused; i.e., while
it is likely that the majority of counts originate from the
object proposed as the identification, the RASSBSC
indicates that at least one fainter source
was present in the uncertainty region. The fainter source
contribute to the signal.
The criterion used for classifying a source as confused
is that the brightest RASSBSC source
must be at least four times brighter than any other source
within the uncertainty
region. Where this is not the case (ie where the RASSBSC
sources differ by less than a factor of four in brightness)
no single identification is
proposed. In this case we
assume that the brightest RASSBSC
source (based on 0.1-2.4 keV ROSAT Position Sensitive Proportional Counter
is also the brightest in the 1-8 keV EXMS range.
bNumber of sources outside the uncertainty region, but within the FOV.
Flags have been assigned to all entries to distinguish the type of identification; they may also be considered as approximate quality flags. This scheme is summarized in Table 2, together with the numbers and percentages of sources in each category. 51% of the entries in the slew survey are identified with confidence - "internal'' sources where the most likely X-ray source also lies within the uncertainty region. A further 29% of entries consist of identified sources which lie outside their formal uncertainty regions. Although these identifications are considered secure, the derived count rates are likely to contain larger errors than the internal sample, due to the error in the initial fit to the lightcurve (this problem is particularly severe for X-ray pulsars where the duty cycle of the pulsations may be 50%). The mean reduced 2 for the internal sample is 1.3, whereas that for the external sources it is 48. The majority of these identifications are X-ray binaries. Hence, 82% of the entries are given single identifications, similar to the identified fraction in the Einstein slew survey. A further 2% of sources are "confused'' - there is a fainter source (at least four times fainter, based on RASSBSC count rates) present in the uncertainty bound.
A further 5% of entries consist of cases where multiple sources of similar brightness lie within the formal uncertainty region. 13% of the entries cannot be identified with any known X-ray emitters, either because there are no X-ray emitters in the vicinity or there is significant doubt about the likely identification. In the case of the 82% of entries where one identification is proposed (sources with Flags of 2, 3, and 4), the count rates given in Table 3 are corrected for collimator transmission. This is achieved by calculating the offset of the source (based on its catalog coordinates) from the centroid of the detector (see line CM in Fig. 7). The correction is proportional to this offset, except for the small "flat-top'' at the center of the FOV (see Fig. 2). In addition, count rates are normalized to count s-1half-1, ensuring consistency with the count rates present in the ME database.
Information on the 992 sources with single identifications is presented in Table 3. The first column is a designation in the form EXMS BHHMM+DDd, where HHMM and DDd are the RA and Decl. of the uncertainty region centroid in epoch 1950 coordinates. (Note: in order to maximise the usefulness of Tables 3, 4 and 5, all coordinates therein are presented for epoch 2000, although all prior catalog searches and computations were performed using epoch 1950 coordinates).
Most detections have a unique EXMS classification. In cases where the classification is not unique, designators in the form (A), (B), etc are appended. In order to usefully group identifications, detections have been ordered in RA of the proposed candidate. Detections with the same coordinate (but unique designator) may be split between tables depending on the category of the identification. In cases where an identified source was observed more than once, the entries are additionally ordered by detection time. The format for the time entries is YYYY/DDD HH:MM, indicating an observation at MM minutes after HH hours on day DDD of year YYYY. The proposed identifications are given (truncated to 18 characters), followed by a "type'' string which gives the first three letters of the catalog from which the identification was drawn (see the list of catalogs above), or the object category. The object categories are XRB: X-ray binaries; SNR: supernova remnant; CLU: cluster of galaxies; and AGN: active galactic nucleus. In this context we adopt the widest possible meaning of the term AGN, including Seyfert galaxies, BL Lacertae objects, QSOs, quasars, Radio Galaxies and Optical Violent Variables (OVVs). The next parameters are the collimator corrected count rate, together with the 2 of the fit to the light curve and the number of degrees of freedom (dof). Large values of 2/dof may indicate source variability. Parameter is the likelihood detection statistic which is the difference in the logarithm of the likelihood between the best-fit and null-hypothesis models with no source present. In the null-hypothesis, is expected to be distributed as 2. Flag values are given under the column labelled "F'', (ranging between 2 and 4 for this sample). Finally, the collimator distances, D, (CM in Fig. 7) are given in arc minutes. Detections with collimator distances 30' may contain large errors in their corrected count rates and should be treated with caution. Remarks on individual entries are given in Sect. 3.5.
The 58 detections with multiple candidates (Flag = 1) are summarized in Table 4. The names of the proposed candidates are stored in a string, the first 20 characters of which are printed in Table 4, together with the number of possible identifications, .The count rates given are raw values, uncorrected for collimator transmission. More information on the proposed candidates is given in Sect. 3.5.
Table 5 contains information on the 160 unidentified detections (Flag = 0). In this case, in addition to the coordinates of the centroid of the uncertainty region, positions of the four bounding corners are also given (in decimal format for brevity), under columns labelled C1 to C4. Again, the count rates are uncorrected for collimator transmission.
EXMS B0016-722 -- AQ Tuc lies within the uncertainty region and is within 10' of 5 X-ray sources. Globular cluster 47 Tuc is nearby.
EXMS B0029-841 -- Four PKS sources lie within the uncertainty region, but none are cataloged X-ray emitters.
EXMS B0110+649 -- The XRB 2S0114+65 is nearby.
EXMS B0113-417 -- UV Phe lies within the uncertainty region, but is not a cataloged X-ray emitter. The RASSBSC source adopted as the identification is consistent with the location of a cataloged X-ray emitting Seyfert galaxy.
EXMS B0114+005 -- 1RXSJ011704.2+000025 may be associated with E0114.4-0015, a Seyfert 1.
EXMS B0121+341 -- NGC 513 lies within the uncertainty region. It is not a cataloged X-ray emitter but is bright and close.
EXMS B0245-420 -- DK Eri lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0250+415 -- Two X-ray counterparts of NGC 1129. Multiple entries are 1H0251+414 and IPC025113+41.
EXMS B0250+417 -- 1H0251+414 and IPC025113+41 are two X-ray counterparts of NGC 1129.
EXMS B0250+418 -- see above.
EXMS B0252-417 -- PKS0252-41 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0252+415 -- an unidentified source possibly associated with the above object. Another unidentified source, EXMS B0252+411, lies less than one degree away.
EXMS B0252+411 -- see above.
EXMS B0309-317 -- PKS0309-31 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0310+411 -- May be NGC 1275 or Mrk 1073.
EXMS B0402-654 -- PKS0403-65 lies within the uncertainty region but is not a cataloged X-ray emitter.
EXMS B0431-613B -- PKS0429-61 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0439-129 -- PKS0436-129 lies within the uncertainty region but is not a cataloged X-ray emitter.
EXMS B0446+447 -- 4C +44.12 and 4C +44.13, cataloged X-ray emitting sources in 3 C 129, lie within the uncertainty region.
EXMS B0525-329 -- MS 05267-3301 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0529-650 -- The proposed RASSBSC identification is near the transient X-ray binary EXO53109-66.
EXMS B0559-664 -- possibility of confusion with A0535-668. Four RASSBSC sources within uncertainty region.
EXMS B0532-664B -- see above.
EXMS B0533-663B -- see above.
EXMS B0533-662A -- see above.
EXMS B0534-657 -- see above.
EXMS B0537-713 -- Probably an LMC object; LMC X-1 and X-2 were both slewed over near the detection time, although neither fell within the uncertainty region, but the raw (and collimator corrected) count rate is too high for either. PKS0531-71 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0601-701 -- an unidentified source less than one degree from another unidentified source, EXMS B0606-697.
EXMS B0606-697 -- see above.
EXMS B0607-712 -- PKS0611-71 lies within the uncertainty region, but is not a cataloged X-ray emitter. An unidentified source, EXMS B0610-714, lies less than one degree away.
EXMS B0610-714 -- see above.
EXMS B0613+228 -- Two obscure radio sources, PKS0615+22 and 4C +22.14, neither cataloged as an X-ray emitter.
EXMS B0613+229 -- 4C +22.14 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0629-713 -- an unidentified source less than one degree from another unidentified source, EXMS B0640-715.
EXMS B0640-715 -- see above.
EXMS B0648-698 -- PKS0650-70 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0654-559 -- PKS0649-55 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0655-707 -- an unidentified source less than one degree from another unidentified source, EXMS B0656-697.
EXMS B0656-697 -- see above.
EXMS B0658+753 -- 4C +74.12 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0821-425A -- 1ES0821-426 is a cataloged X-ray source.
EXMS B0821-425B -- an unidentified source which might be associated with the above, although it does not lie within the uncertainty region.
EXMS B0820-424 -- see above.
EXMS B0823-426 -- see above.
EXMS B0834-428 -- probably G0834-430 and/or 4U0836-429; both were slewed over near the detection time though neither falls within the uncertainty region.
EXMS B0834+254 -- Two extragalactic objects, Mr k1218 and B20834+25 lie within the uncertainty region but only Mr k1218 is a cataloged X-ray emitter.
EXMS B0912+354 -- 4C +34.31 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0917-549 -- PKS0916-54 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B0923-308 -- PKS0923-30 lies within the uncertainty region, but is not a cataloged X-ray emitter. The proposed RASSBSC identification is near X-ray source 1H0919-312.
EXMS B1042-599 -- Eta Carinae lies within the uncertainty region, but is inconsistent with the position of the RASSBSC source.
EXMS B1044-595 -- see above.
EXMS B1043-593 -- see above.
EXMS B1040-593 -- see above.
EXMS B1049+385 -- B21049+38, a high-redshift radio galaxy (or possibly Seyfert 2), lies within the uncertainty region but is not a cataloged X-ray emitter.
EXMS B1123-588 -- The proposed RASSBSC identification may be associated with the X-ray emitting SNR MSH11-54.
EXMS B1153+317 -- 4C +31.38 lies within the uncertainty region but is not a cataloged X-ray emitter.
EXMS B1213+038 -- The uncertainty region contains several radio sources, one of which (4C +04.41) corresponds to X-ray source 1ES1215+039.
EXMS B1155-187 -- Three obscure emission line galaxies lie within the uncertainty region.
EXMS B1235+708 -- 4C +70.13 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B1246-410B -- ESO 323-G32 is a Seyfert 2 in cluster A3526.
EXMS B1254+276 -- multiple unclassified AGN in Veron catalog. Multiple entries all have name UNKNOWN.
EXMS B1324-312 -- A3558 is a cataloged X-ray emitter.
EXMS B1415+253 -- NGC 5548 and 1E14156+259 are two X-ray emitting AGN.
EXMS B1415+255 -- see above.
EXMS B1416+256 -- see above.
EXMS B1517-613 -- TrA X-1 was slewed over near the detection time but would have appeared near the edge of the wrong quadrant.
EXMS B1525+525 -- 4C +52.35 lies within the uncertainty region but is not a cataloged X-ray emitter.
EXMS B1550-609 -- 4U1543-624 and 4U1556-605 were both slewed over near the detection time, but neither fell within the uncertainty region.
EXMS B1611-508 -- HW Nor is within 10' of 18 X-ray sources.
EXMS B1637-671 -- 4U1627-673 was slewed over near the time of detection but would have appeared in the wrong quadrant.
EXMS B1658-228A -- an unidentified source less than a degree from another unidentified source, EXMS B1658-228B.
EXMS B1658-228B -- see above.
EXMS B1702-429 -- probably 4U1705-440 and/or 4U1702-429; both were slewed over near the detection time though neither falls within the uncertainty region.
EXMS B1719-436 -- an unidentified source less than a degree from another unidentified source, EXMS B1722-442.
EXMS B1722-442 -- see above.
EXMS B1721-231A -- an unidentified source less than a degree from two other unidentified sources, EXMS B1721-231B and EXMS B1721-231C.
EXMS B1721-231B -- see above.
EXMS B1721-231C -- see above.
EXMS B1730-445 -- 4U1735-444 passed slightly beyond the field of view near the time of the detection.
EXMS B1734-155 -- TX Ser lies within the uncertainty region and is within 10' of 1RXSJ173735.1-152357 and 2E1734.7-1522.
EXMS B1741-337 -- PKS1742-337 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B1743-300A -- probably GC X-1 or one of the many nearby objects, most of which were slewed over but none of which fell within the uncertainty region.
EXMS B1743-300B -- see above.
EXMS B1743-303B -- probably one or more of the many Galactic Center objects including SL1744-300 and neighboring sources, many of which were slewed over but none of which fell within the uncertainty region.
EXMS B1747-366 -- probably 4U1746-360 and/or A1744-361; both were slewed over near the detection time though neither falls within the uncertainty region.
EXMS B1754-289 -- an unidentified source less than a degree from another unidentified source, EXMS B1755-295.
EXMS B1755-295 -- see above.
EXMS B1827-100 -- an unidentified source less than a degree from another unidentified source, EXMS B1828-100.
EXMS B1828-100 -- see above.
EXMS B1837-052 -- Near the SNR G27.4+0.0.
EXMS B1846-029A -- In addition to EXO1846-03, G1845-03 and A1845-024 were also slewed over near the detection time but neither fell within the uncertainty region.
EXMS B1905+009 -- Probably 4U1905+000 and/or Aql X-1, both of which were slewed over but neither of which fell within the uncertainty region. V810 Aql does lie within the uncertainty region and within 10' of five Rosat sources.
EXMS B1906+007 -- 4C +00.71 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B1931+800 -- 4C +79.20 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B1948+113 -- 4C +11.59 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B1955+123 -- an unidentified source less than a degree from 4U1957+115, and also another unidentified source, EXMS B1955+124.
EXMS B1955+124 -- see above.
EXMS B2124+751 -- 4C +74.27 lies within uncertainty region, but is not a cataloged X-ray emitter.
EXMS B2216-389 -- Q2217-391, an extragalactic object, lies within the uncertainty region, but is not a cataloged X-ray emitter. 1H2217-392 is a nearby stellar X-ray source.
EXMS B2232-734 -- PKS2238-73 lies within the uncertainty region but is not a cataloged X-ray emitter.
EXMS B2233-657 -- PKS2239-65 lies within the uncertainty region, but is not a cataloged X-ray emitter.
EXMS B2223-331 -- PKS2224-33 lies within the uncertainty region, but is not a cataloged X-ray emitter.
|Figure 10: The histogram of observed source offsets (in 60 bins) for the 463 sources identified as X-ray binaries whose positions are within the formal uncertainty regions|
|Figure 11: A simulated histogram of source offsets (in 60 bins) for 500 sources, produced by averaging ten simulations. The variance in each bin is shown by a vertical line. The observed histogram (Fig. 10) is plotted using a broken line|
The offset of a given detection in the EXMS is simply the angular distance between the centroid of the uncertainty region and the catalog coordinates of the proposed identification. Since it is required that all identifications passed through the FOV, the maximum possible offset is given by the 45' FWHM of the collimator response. For simplicity the offsets are broken down into two components, defined in Fig. 7. The absolute offset for source E is the scalar distance EC and the vector offset is EM. Similar offsets are defined for internal sources.
|Figure 12: Distribution of vector offsets (component of offset perpendicular to the direction of slew; vector EM in Fig. 7) for the 463 sources indentified with X-ray binaries which also lie within the formal uncertainty regions|
|Figure 13: Plot of absolute offsets (the distance EC in Fig. 7) against time since start of first slew leg, for the 463 sources indentified with X-ray binaries which also lie within their own error boxes. No time-dependent effects are evident|
|Figure 14: Plot of vector offsets against time since start of first slew leg, for the same sample as in Fig. 12. Again no time-dependent effects are evident|
|Figure 15: Plot of vector offsets against galactic longitude, ,demonstrating the lack of any dependence|
Absolute offsets were derived for the 463 XRBCAT identifications in the EXMS, and these were then summed into bins of 60, resulting in the histogram of Fig. 10. In order to understand the shape of this distribution, a simulation was performed in the following manner. First, a population of X-ray binaries was created with a realistic count rate distribution. This was achieved by correlating the XRBCAT against the ME database to obtain normalised ME counts for each matching source. Sources were then drawn randomly from this population and assigned random positions along a line, approximating the narrow uncertainty regions of the true sample. The observed counts of each source were then adjusted in accordance with the distance of the source from the center of the line, using the ME collimator response function. The average background count rate of the 1-8 keV lightcurves is 15 countss-1half-1, and this was adopted in the simulations. Since the background events are predominantly particle-induced, they are independent of collimator distance. As a consequence, faint sources are not detected unless they fall near the middle of the simulated FOV, but as such sources are more numerous, they play an important role in the shape of the offset histogram. The procedure was repeated until 460 sources had been detected. Since there are <100 detections in each bin, the detailed shape of the histogram is dependent on statistical fluctuations. In order to characterize the shape and variance of a mean histogram, the results of ten independent simulations were averaged. The overall shape of the observed histogram is reproduced satisfactorily, with an approximately flat-topped distribution out to about 0.6 of the maximum offset (Fig. 11). In most bins, the real and simulated histograms agree within the variance of the simulation, and the small discrepancies that exist can be understood in terms of limitations in the simulation, particularly in the manner in which the detection threshold is modeled, and the fact that a few bright sources contribute predominantly to the detected numbers of sources in the high-offset bins.
Having established that the source offsets for this subsample are free of obvious systematic effects, the behavior of the vector component of the offsets which lie in the slew direction was examined. In the absence of systematic errors in the pointing and timing information, the resultant distribution of offsets should be clustered symmetrically around zero. This is indeed the case, as shown in Fig. 12, where the 463 vector offsets are plotted against raw count rate. The mean vector offset is consistent with zero: '. As is to be expected, the spread in the offsets is largest for the faintest sources.
|Figure 16: As for Fig. 15, but for galactic latitude|
|Figure 17: EXMS collimator corrected count rates for the Cas-A SNR (assuming a point source). The expected count rate is 108 counts-1half-1. With the exception of the measurement on 1984/201, which occured at an offset of 35', the count rates lie within 15% of the expected value|
|Figure 18: Collimator corrected count rates for the Tycho SNR (assuming a point source). The expected count rate is 32 counts-1half-1, consistent with the mean EXMS value of counts-1half-1|
|Figure 19: Curves for converting from EXOSAT ME count rates to deadtime corrected fluxes for power-law spectra. From top to bottom, the curves give the predicted count rates for a spectral normalization of unity for values of <1021, 1021, ,, 1022, ,1023, , and atoms cm^-2atoms cm-2|
The behavior of the absolute and vector offsets as a function of the elapsed time since the start of each slew was also examined, to ascertain whether the spacecraft attitude reconstruction gradually becomes less accurate. Since the slew durations vary from hundreds to thousands of seconds, the data span a wide time range, but it is clear (see Figs. 13 and 14) that the offsets appear random and independent of elapsed time.
Finally, the distributions of the absolute and vector offsets as a function of position in the sky were examined. Due to the sample being drawn solely from X-ray binaries, the sources are clustered around the Galactic plane and the Magellanic clouds. This means that there are large areas of the sky where the offsets are not sampled. However, at least for the subsample, there is no suggestion of any position-dependent offset effects in either equatorial or Galactic coordinates (see Figs. 15 and 16 for the latter).
Due to the variability of most X-ray sources, there are few sources in the EXMS which enable the accuracy of the derived count rates to be estimated. The brightest sources, such as X-ray binaries, are highly variable, while those sources which are time-invariant tend to be faint and extended (the latter leading to incorrectly estimated collimator corrections). Nevertheless, the observations of SNRs within the EXMS are in good agreement with the expected count rates based on ME pointed observations.
As an example, in 12 detections of Cas-A between 1983 and 1986, the derived EXMS count rates vary between 97 and 180 counts-1half-1. Most values are between 100 and 120 counts-1half-1, with a mean value of 114 9 counts-1half-1 (see Fig. 17). The detection with the highest count rate is also that with the largest offset distance of 35' (see Table 3). The extended nature of the source leads to an overestimated collimator correction. However, the mean value is in good agreement with the count rate during the single pointed ME observation of Cas-A of 108 counts-1half-1. Similar agreement is seen for the 15 slew detections of Tycho, with a mean value of 31 5 counts-1half-1, compared with 32 counts-1half-1 from two pointed observations (Fig. 18).
|Figure 20: EXMS detections of the XRB Ser X-1. Two pointed observations performed on 1985/251 and 252 recorded count rates of 344 and 343 counts-1half-1, similar to the EXMS values|
|Figure 21: EXMS detections of the XRB 4U0918-549 showing an apparant long-term decrease in 1-8 keV intensity|
|Figure 22: EXMS detections of the XRB LMC X-3. The 198 (or possibly 99) day periodicity discovered by Cowley et al. (1991), may be evident|
Copyright The European Southern Observatory (ESO)