The search for potential short-period variables was conducted
by analyzing light curves included in data bases for the B and
V filters.
To select potential variables we employed two methods
which are described in details by Kaluzny et al. (1996).
The first method is based on selecting stars exhibiting ``noisy'' light
curves. An 
test is applied to the light curves to
identify such stars. In the second method the light curves are scanned
with a filter designed to detect an eclipse-like events.
To illustrate the overall quality of photometry we present in Fig. 3 (click here)
a plot of rms deviation versus the average V magnitude for the light
curves containing at least 60 data points.
Seven certain variables were identified this way.
The data collected on the night of 8/9 July 95 are of superior quality
as compared with the remaining nights. Therefore, these data were
subject to an additional analysis. As a result one more variable
(star V11 in Table 2 (click here)) was identified. This variable
shows a light curve with a very low amplitude and it is located
very close to the projected cluster center.
The coordinates and basic photometric properties of the identified variables are listed in Table 2 (click here). The listed rectangular coordinates correspond to the positions of variables on the ``template" V-band image which was submitted to editors of A&A (see Appendix A). The equatorial coordinates are derived from reference to 265 HST guide stars (Lasker et al. 1988) identified in the surveyed field. The adopted frame solution reproduces equatorial coordinates of these stars with residuals not exceeding 0.7 arcsec.
Table 2:
Rectangular coordinates and basic photometric data for variables V3-12
from the field of NGC 6397. Uncertain quantities are marked with a colon.
The periods are given in days and 
is a full amplitude
of variability in the V-band
The periods of pulsating variables were determined using the aov statistic (Schwarzenberg-Czerny 1989, 1991). Several moments of individual minima were obtained for eclipsing binaries V4-9 using the method of Kwee-Van Woerden (1956). These moments are listed in the second column of Table 3 (click here). The third column of that table contains formal errors of the minima. Because of the short time-base of our observations all periods listed in Table 2 (click here) are of low accuracy. In particular, the period of V6 is just a rough estimate derived from an incomplete light curve. However, we exclude a possibility of any gross errors caused by aliasing related problems.
Variable V3 is listed in the Hogg (1973) catalogue as a
background object with the period 
.
We revised the period of V3 to 
.
Variable V10 is most probably identical to the variable blue straggler
BSS #533 reported by Rubenstein & Bailyn (1993).
This conclusion is based on a close similarity of periods, light curve
amplitudes and magnitudes reported for V10 and BSS #533.
Variables V4-9 and V11 are, to the best of our knowledge, new
discoveries
Figure 3: The single-measurement errors of our photometry
versus the average V magnitudes for 16177 stars whose light curves were
examined for variability
Table 3: Moments of minimum light determined from the V-band light
curves of variables V4-9
The phased light curves of variables V3-4, V7-9 and V11 are shown in Fig. 4 (click here). A time-domain light curves of V5, V6 and V10 are presented in Fig. 5 (click here). Only the V-band light curves are displayed in this paper. The tables containing B and V light curves for all the variables discussed here are available from A&A (see Appendix A).