Observations of social play in several species of mammals indicate that individuals often depart from the usual social conventions by, for example, alternating dominant and subordinate positions in ways that do not occur outside the play context. Some researchers have even suggested that individuals must follow a 50:50 rule during dyadic play (play between two individuals), so that each participant wins an equal proportion of play encounters. Commonly cited cooperative tactics used to equalize play include self-handicapping (participants make themselves more vulnerable to attacks by their opponents) and role reversal (individuals that are dominant in the nonplay context appear subordinate during play). Where such tactics occur among unevenly matched opponents, they appear to facilitate play by making play more appealing to the less advantaged player. When Bauer and Smuts set out to study play behavior in domestic dogs, they made several predictions. They expected to find no significant sex differences in dogs' play behavior. The motor skills dogs use in play fighting parallel those used in nonplayful aggression and hunting, areas in which dogs' behavior is relatively undifferentiated by sex. They also predicted that the advantages imparted by larger relative size, by the experience of age, and by higher dominance status would affect dogs' dyadic play. Existing research on a variety of animal species suggests that individuals with such advantages often refrain from attacks and pursuits or engage in self -handicapping so as not to intimidate their play partners. If this held t rue for dogs, larger, older, more dominant dogs would show more self-handicapping than their partners. But Bauer and Smuts predicted instead that dogs would deviate from the hypothetical 50:50 rule, with advantaged individuals retaining their positions by performing the majority of attacks and pursuits and engaging in fewer self-handicapping behaviors than their partners, thus reinforcing existing hierarchies. Bauer and Smuts' three-year study of dogs' dyadic play found that most dyads showed some degree of asymmetry (one dog winning more encounters than the other) and some dyads showed complete asymmetry. They also found that in general, older dogs performed more attacks and pursuits and that younger dogs engaged in more self-handicapping. Role reversal between dominant and subordinate individuals varied widely: several dyads never reversed dominance roles, a few reversed them frequently, and most reversed them occasionally. Bauer and Smuts' finding about asymmetry in dyadic play has several implications. First, it indicates that active self -handicapping and role reversals are not necessarily required for play to occur. Indeed, play often continued at length even when one partner always won. Second, since frequent role reversals occurred, it appears that normal status asymmetries are often significantly more relaxed in the play context. This suggests that role reversals, while not always necessary, probably do facilitate play.