We used two sets of videotaped data of playing domestic dog dyads to determine whether rolling over during play served as a signal of submission or whether it was a combat maneuver adopted as part of an ongoing play sequence. Our results provide strong support for the latter. In the absence of any overt indication of agonism, the frequency with which rollovers occurred was determined primarily by play bout length. The discrepancy in partner size had no effect on the probability that rollovers would occur and there was no evidence that smaller dogs were more likely to rollover or to sustain a supine posture for longer, if they did. The supine phase of rollovers was significantly skewed to short durations. Most rollovers were either defensive (evading a nape bite) or offensive (launching an attack). None could be categorized as submissive. We conclude that asymmetries in the performance of rollovers cannot be assumed to point to asymmetries in the relationships between play partners.
This article is part of a Special Issue entitled: Canine Behavior.