The first phase of commissioning involved laboratory tests to determine
various parameters like linearity, gain and readnoise of the detector, vignetting in the optics etc.
Measurements were also made with a 3000 K filament source to estimate
depolarization and instrumental polarization in 
bands, and a wide band defined by a 3 mm thick KG3 glass. A 100% polarized
beam showed that the depolarization is about 0.4% in the B and
V bands, less than 1% in the R band, and less than 6% in the broad band.
There was no discernible rotation of the position angle between these bands.
For R and the wideband, the accuracy of measurements
is limited by the performance of the polarizer used. Measurements made with
a Lyot depolarizer introduced in front of the artificial star showed that
the instrumental polarization is about 0.03% in the
B-band and less than 0.06% with the wideband filter.
The second stage of tests were conducted when the instrument saw the
first light in February 1996, at the
1.2 m Gurushikhar
Infra-Red Telescope facility
on Mt. Abu, Rajasthan, operated by the Physical Research Laboratory at
Ahmedabad, India.
Table 2 (click here) contains the results of some of the observations of
standard stars in the V-band.
Column 1 of
the table gives the identification number of the star. Columns 2-5 contain
the measured values and
errors of the fractional polarization p and the position angle
. Column 6 gives the error 
based on
photon statistics alone, derived as illustrated in Sect. 3. Columns 7 and 8
list the published values of p and .
The first two stars in the table are polarized standards while the last two are unpolarized ones. The four measurements of star HD 39587 are made at different corners of the CCD frame in order to verify the uniformity of response. Comparing the measured and published results for the polarized stars we see that there is no indication of depolarization.
| Star | p |
| | 
|  | p0 |  |
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) |
| HD | % | % |  | | % | % |
|
| 43384 | 3.05 | 0.04 | 172 | 0.3 | 0.03 | 3.0 | 170 |
| 154445 | 3.65 | 0.03 | 90.5 | 0.2 | 0.02 | 3.7 | 90 |
| 102870 | 0.03 | 0.01 | 64.4 | 7.8 | 0.02 | 0.017 | 162 |
| 39587 | 0.03 | 0.03 | 140 | 27.5 | 0.02 | 0.013 | 20 |
| 0.05 | 0.02 | 159 | 12.7 | 0.02 | |||
| 0.09 | 0.02 | 34 | 6.7 | 0.02 | |||
| 0.05 | 0.02 | 167 | 11.3 | 0.02 | |||
|
|
Neither is there any rotation of the position angle beyond a couple of degrees. From Col. 3 and Col. 6 for the unpolarized stars, we see that the error in the measured polarization is comparable with that expected on the basis of photon statistics alone.
In Fig. 4 (click here), the values of
are plotted against 
for all the unpolarized
standards observed in the V-band. The extremely low correlation coefficient
between q and u, is indicative of the low instrumental
polarization.
The average values of q and u, give a value of polarization
. These observations show that for accuracies 
0.05%
in the measurement of polarization p, the
performance is still limited by photon noise and the instrument polarization
floor has not yet reached.
Figure 4: Normalized Stoke's parameter q is plotted against u for a
number of unpolarized standard stars. The correlation coefficient of the
points is about 0.06 and the average values of q and u give a
value of