The hitherto known population of pre-main sequence stars in Lupus, as found by optical surveys, is concentrated in four subgroups (Krautter 1991). One of the main goals of our survey of TTS in the Lupus SFR was to determine the spatial distribution of these stars on a relatively large scale, to see whether this clustering also applies to X-ray selected TTS. The spatial distribution of TTS as found by our work is shown in Fig. 3 (click here). TTS found by means of pointed ROSAT observations are not plotted, because these pointed observations are biased spatially. For comparison we show the spatial distribution of the previously known TTS. It is immediately obvious that our sample of X-ray selected TTS in this SFR is spatially distributed over the whole region of interest, in marked contrast to the TTS known prior to ROSAT.
Figure 3: Spatial distribution of our sample of new RASS-discovered TTS
(left panel) and the previously known TTS in the study region (right panel).
CTTS are mrked by filled squares, WTTS by open squares.
Gould's Belt is shown by the solid line. Also plotted are the two
lowest contours of the CO map of Murphy et al. (1986)
Following our paper on a similar survey in the Taurus-Auriga SFR (Wichmann et al. 1996), we applied two different statistical tests, the nearest-neighbour distance (NND) test (Gomez et al. 1993) and the two-dimensional Kolmogorov-Smirnov (D2KS-) test (Fasano & Franceschini 1987; Press et al. 1992) to our sample of newly discovered TTS in order to obtain quantitative information on their spatial distribution and its difference with respect the TTS known prior to ROSAT. As the pointed observations have been carried out in the areas where the previously known TTS are clumped, TTS found by means of these pointed observations are not used for this analysis.
The distribution of nearest-neighbour distances for the RASS-discovered TTS
as well as for the TTS known prior to ROSAT is shown in Fig. 4 (click here). As
expected, the previously known TTS show strong clustering, with a median
NND of 
, much lower than the expectation
value of 
, for a uniform random
distribution. For the RASS-discovered TTS we obtain 
as compared to the expectation value of 
.
The significance of the observed excess of small NNDs for the RASS-discovered
WTTS can be calculated in the following way:
in a given bin, the probability for n or more counts when
counts are expected, is given by the
cumulative Poisson probability
. A random sample will yield the observed excess
for 
with a probability of 
;
for 
with a probability of 
;
for 
with a probability of 0.248;
and for 
with a probability of 0.011.
Thus at the smallest NNDs, the observed excess is indeed highly significant. From Fig. 3 (click here) it seems that in fact some clumps may be present in the distribution, along with a possible overdensity in the northern part of the study area.
This conclusion is further strengthened by the result of a two-dimensional
Kolmogorov-Smirnov test (for a description see Wichmann et al. 1996).
Comparing the spatial distribution of the RASS-discovered TTS to a model of
uniform distribution, the test statistic D has a value of
D = 0.283, corresponding to a probability P of only P = 0.0005 for the
hypotheses of a uniform distribution.
A closer inspection shows that
the most significant dividing point
is 
and 
,
with the quadrant southeast of this point being
the most underpopulated with respect to a uniform distribution.
A comparison with Fig. 3 (click here) shows that
in fact many of the new WTTS seem to be located around Gould's Belt
(cf. Gould 1879; Bahcall et al. 1987),
while especially the southeastern part of our survey area, i.e. the part
most distant from Gould's Belt, seems to show a low
surface density of new WTTS.
As the galactic plane lies south of our survey region
(in Fig. 3 (click here), the southeastern edge is at 
, the
southwest edge at 
)
this implies that the spatial distribution of RASS-discovered WTTS in
Lupus correlates with Gould's Belt rather than with the galactic plane.