One of the most significant and stubborn mysteries in astrophysics today concerns the nature of the dark matter in spiral galaxies (e.g.~Rubin et al.~1978). The least radical candidate for such dark matter is baryonic objects which are too large to be detected as dust or gas. Indeed, if the Hubble constant is not too large, a significant fraction of the baryons, as implied by Big Bang nucleosynthesis, must be hidden as dark matter (Walker et al.~1991). Arguments have been made for why such dark baryonic objects are unlikely on individual mass scales of atoms to brown dwarfs (Hills 1986, Hegyi \& Olive 1986). Still, objects of primordial composition and more massive than about $10^{-7} M_{\odot}$ might be expected to resist evaporation until the present day (de R\'ujula et al.~1992), while masses smaller than about $0.077 ~ M_\odot$ would fail to ignite as stars (Burrows et al.~1993).

Astrophysicists' frustration explaining the dark matter with any directly detectable objects has led to the suggestion that gravitational microlensing might be used to at least betray the presence of individual objects via their effects on background stars as sources (Paczynski 1986), and thereby give some indication as to their mass. Such searches have recently taken place towards the Large Magellanic Cloud (LMC) (Alcock et al.~1996, Aubourg et al.~1995) and Bulge (Alcock et al.~1995, Paczynski et al.~1994), with searches towards the LMC ruling out most of the dar k matter being composed of substellar-mass objects (Aubourg et al.~1995, Alcock et al.~1996) heavier than about $10^{-6} M_\odot$, while suggesting that a large fraction might have the same component mass as low-mass stars (Alcock et al.~1996). Given the uncertainty of the Galactic halo's distribution of MAssive, Compact, Halo Objects (MACHOs) and therefore the lensing geometry leading to events, the relationship between mass and observed microlensing lightcurve timescale is still unclear.

In part because of its unique geometry with respect to Earth and partially due to high predicted optical depths ($\tau$) due to lensing, M31 is a uniquely powerful venue for studying microlensing. Early we realized that an M31 microlensing survey would show particular advantages if the practical aspects of studying such a distant, crowded field of stars could be overcome.

We found such an approach, briefly outlined by Crotts (1992) with a complete description of the realistic technique and preliminary results found in Tomaney and Crotts (1996, hereafter TC). By subtracting images in a time sequence, then performing difference image photometry'' (DIP, also know as pixel lensing''), we can study the residual point sources due to variables, while the signals from the many crowded, non-varying stars subtract away. With a practical method of observation and analysis, we can exploit the advantages inherent in studying M31: 1) very small component mass limits, due to the small angle subtended by the photosphere of M31 stars compared to the Einstein radius of objects of solar mass (c.f. TC for low-mass results), 2) the ability to study different parts of M31, thereby studying the spatial distribution of microlensing objects, 3) the ability to study many stars at once in fields of high $\tau$, thereby detecting events in short periods of observation, and 4) the constrained microlensing geometry, due to the fact that lensing mass is concentrated over the center of the galaxy, thereby allowing a better determination of the MACHO mass given microlensing event timescale.

It is our hope that by studying M31 in this way, both its halo and bulge, one can more readily understand both these results and those obtained in the Galaxy. This paper presents our results from our first season of observation toward this goal.