We model the disassembly of an excited nuclear system formed as a result of a heavy ion collision. We find that, as the beam energy in central collisions in varied, the dissociating system crosses a liquidgas coexistence curve, resulting in a first-order phase transition. Accessible experimental signatures are identified: a peak in the specific heat, a power-law yield for composites, and a maximum in the second moment of the yield distribution. [S0031-9007(97) PACS numbers: 25.70. Pq, 21.65. + f, 24.10.Pa, 64.60.My Nuclear matter is a fictitious arbitrarily large N Z system in which the Coulomb interaction is switched off. Mean-field theory for nuclear matter has been applied many times, and it is well known that such a system shows a van-der-Waals-type liquid-gas phase transition. It was suggested in the early 1980s [1,2] that in heavy ion collisions at intermediate energies one might be able to probe this liquid-gas phase transition region. In heavy ion collisions, matter would be heated as well as compressed. This compressed blob would then expand passing through the liquid-gas coexistence phase. One might be able to extract information about this region from selected experimental data. Earlier, the Purdue group had conjectured that the breakup of large nuclei by energetic protons would show signatures of critical phenomena [2].There are several complications which make the study of phase transitions in nuclei difficult. The systems are small, thus singularities get replaced by broad peaks. The collision is over quickly. The existence of thermal equilibrium has sometimes been questioned and replaced by quite complicated transport equation approaches. Often one has been content to do calculations where the objective is to fit the experimental data. Once this is achieved the question of a possible liquid-gas phase transition is not addressed. There are models where such questions are irrelevant, or at least very indirect, such as various models based on sequential decays. The literature on calculations for fragment yields in intermediate energy heavy ion collisions is huge. We will not attempt to mention all approaches.In this paper we will focus on the liquid-gas phase transition using a lattice gas model [3]. Previously we have used the model to fit data on central collisions [3], on peripheral collisions [4], and for central Au on Au collisions [5]. The model gives a fair description of data in those instances-does it say anything definite about the experimental signature of liquid-gas phase transitions?In the lattice gas model we place n nucleons in N cubes, where n is the number of nucleons in the disassembling system, and N͞n r 0 ͞r f , where r 0 is the normal nuclear density, and the disassembly is to be calculated at r f , the "freeze-out" density beyond which nucleons are too far apart to interact. An attractive nearestneighbor interaction is assumed. We place the n nucleons in N cubes by Monte Carlo sampling using the Metropolis algorithm. Once the nucleons have been placed we also ascribe to each...