A total of 602 stars with 
larger than 10 was considered as
possible candidates. The light curves of these stars were
individually inspected and a variety of variable stars and possible
variables was found. To identify the Cepheid variables in the sample,
we searched for stars with light curves similar in appearance to known
Galactic and LMC Cepheids. Of the 602 significantly variable stars,
228 were identified as Cepheid candidates. The remaining variable
stars did not have light curves clearly similar to Cepheid light
curves: e.g., short period variables with periods < 1 day or erratic
variables with only a small number of points far from the mean. Rough
estimates of their periods were made by eye. We did not attempt to
fit model light curves rigorously, partly because of the poor
sampling, but also because exact periods are not crucial to our goals.
We also determined the V magnitude at maximum light, 
, for
each candidate. The error on varied considerably, from
0.02 mag to 0.2, depending on whether the peak of the light
curve was observed or not.
Figure 2: Histogram of period ratios for M 31 Cepheids
observed by us and by Baade & Swope (1962, 1964) or
Gaposhkin (1963)
Figure 3: Comparison of peak magnitudes for M 31 Cepheids
observed by us and by Baade & Swope (1962, 1964) or
Gaposhkin (1963)
Of the 228 Cepheid candidates, 97 were previously identified by Baade
& Swope (1962, 1964) and Gaposhkin (1963), particularly in our
south-most field. Because they have a much larger amount of data
covering a larger baseline, their period determinations are far more
accurate than ours. Figure 2 (click here) shows a histogram of the
ratio between their measurement of the periods and ours for each star
found in common. The central peak has a 1
width of about
15%. A peak can also be seen at 0.5 due to aliasing affecting our
determination. Most of the error in our period determination is due
to the poor sampling, since the full period is not covered for many of
the variables. Figure 3 (click here) shows a comparison of the
magnitude at maximum light for the same Cepheids, as determined by us
and by Baade & Swope (1962, 1964) and Gaposhkin (1963). The
general correlation is clear, though a significant offset can be seen. This
offset is due to the fact that the magnitudes reported by Baade and
collaborators are photoelectric magnitudes, which are close to B,
while ours are V magnitudes. The offset of 0.7 magnitudes
is consistent with the typical colors of these stars.
Figure 4: Period -- Luminosity (PL) diagram. The open
squares show the measurements of M 31 Cepheids from Baade's group
(Baade & Swope 1962, 1964; Gaposhkin
1963). The filled squares show our new Cepheid candidates. The
solid line is the PL relationship determined for a set of Baade M 31
Cepheids by Freedman & Madore (1990)
In Fig. 4 (click here) we present a period-luminosity relationship for the
Cepheid candidates we have identified, along with those identified by
Baade & Swope (1962, 1964) and Gaposhkin (1963). To bring the
Baade magnitudes in line with the V magnitudes, we have simply added the
0.7 magnitude offset, which is sufficiently accurate for our purposes
here. Included in this diagram is a line representing the position of
the PL relationship found by Freedman & Madore (1990) using
improved observations of the Baade & Swope (1964) Cepheids,
from their Field IV, where the extinction is insignificant. It is clear
that these variables fall generally along the P-L relationship, lending
credence to their identification as Cepheids. One variable which was
identified by the light curve turned out to fall far above the P-L
relationship and has been rejected as a true Cepheid. This variable
is probably of Galactic origin, possibly a binary system in the
Galactic halo. There is a small tendency for the points in
Fig. 4 (click here) to fall below the line for 
, most
likely due to the range of extinction for the sample. For 
, there is a small tendancy for the points to land above the line.
This may be due to Malmquist's bias, to the second track due to
Cepheids oscillating at the overtone (e.g., Böhm-Vitense
1994), or because of tendancy for us to underestimate the period (see
Fig. 2 (click here)).
We present lightcurves of all 130 new Cepheid candidates in
Figs. 5-10 sorted in order of their
Right Ascension. Our estimates for the upper and lower limits on
are shown in these light curves by two dashed lines. We
also tabulate observed quantities of these candidates in
Table 1. The candidate rejected on the basis of its
location in the P-L diagram is not included in the light curves or
table, nor are those Cepheids which were already identified by Baade
& Swope (1962, 1964) or Gaposhkin (1963). An estimate of the
contamination of our sample can be obtained from the number of
non-Cepheid variables identified by both Baade's group and our
analysis, which we mis-identified as a Cepheid. This number is very
low: Of 97 objects which we considered to be Cepheids, none were found
to be another kind of variable by Baade's group. This suggests that
our contamination is < 1%, but a more conservative estimate might
be 3%, because of the low number statistics. The completeness can be
estimated in a similar way, by comparing the number of Cepheids which
were found in our fields by Baade's group, but which we failed to
find. By doing this comparison, we find that our completeness is
53%, while the completeness of the Baade dataset is 88%. The
magnitude distributions of both samples are similar, and suggest
significant lack of completeness due to the magnitude limit for
.
Figure 1 (click here) shows the distribution of our newly identified
Cepheid variables along with those identified by previous researchers
(Baade & Swope 1962, 1964; Gaposhkin 1963). The reader is
cautioned that some of the large-scale structure apparent in this image is
due to the limited coverage. The small inset shows the coverage of both
our observations and those of Baade & Swope (1962, 1964) and
Gaposhkin (1963). We have also included in this figure a contour map
from the HI survey of Unwin (1980). This allows the reader
to compare the current location of active star formation with the
location of the Cepheids, which have a typical age of roughly 
. In a companion paper, we will discuss in detail the implications
the observed distribution has for the star formation history.