The concept of an auroral substorm intrinsically involves a large increase in auroral power within a relatively short time (a large dP/dt). There is currently no standard for just how large a power increase is needed to identify a substorm. It is unclear whether auroral brightenings and fadings form a continuous distribution, or even whether large decreases in auroral power also occur within a short time (i.e., a negative, or inverse substorm). We used Polar UVI images of global auroral power to investigate these and related questions. Specifically we considered the distribution of dP/dt and (dP/dt)/P, that is, the distribution of absolute and relative changes in auroral power. At small values of ∣dP/dt∣, negative changes are much more frequent than positive changes. In fact, a small decrease in auroral power is the most frequent change between two consecutive Polar UVI images. Hence the power in the auroral oval is, the majority of the time, in slow decline. Large magnitude changes are rare, but turn out to be almost exclusively positive, implying inverse substorms do not exist. Beyond a 0.2%/s rate of change in auroral power (which amounts to a 37% change over the typical Polar UVI image spacing of 184 s), only positive events occur, within the measurable noise levels. However no clear boundary divides substorms from other types of auroral brightenings: rather the spectrum of large positive changes in auroral power is continuous. These results are relatively insensitive to the exact value of Δt (from 36 s to 6 min), and to whether the premidnight or postmidnight auroral oval is considered.