THIS paper seeks to explain certain observations pertaining to slavery. Several of the implications will be tested against data provided in Fogel and Engerman's recent work on the subject,' in conjunction with a scattering of information from numerous other sources. In a nonslave economy, an individual determines the amount of labor he is willing to supply simultaneously with his consumption decisions. In this respect, labor differs from any other factor of production. In a slave economy, however, both the amount of labor to be extracted from a slave and his consumption or "feeding" are determined by the slave owner. Thus, the labor market for slaves differs from that for freemen by the decision-making unit through which options are exercised. In what follows I postulate that in his labor-supply decision a freeman will take into account nonpecuniary dimensions of his work and of his leisure time. On the other hand, the slaveowner is assumed to act as a wealth maximizer in deciding what amount of labor to demand from his slaves. Thus we abstract from any nonpecuniary aspects in master-slave relationships. At the same time, we recognize that while the slave is subject to the dictates and enforcement of his owner's will, his action also depends on his own preferences for pecuniary as well as nonpecuniary goods. Under these postulates, how would slaves be treated? How would their performance differ from that of free workers, and from that of such other factors of production as machines? The following section compares the work effort and consumption levels of slaves with those of freemen, largely under an assumption of zero policing