PERT Compared with CPM When most of the activities in a project are similar to other activities that have been performed a large number of times, CPM scheduling is generally used. With CPM, estimates of activity durations are based on historical data and are assumed to be the mean or average time that the activity has taken to perform in the past. However, when a project contains a majority of activities for which no experience exists (that is, when no historical data is available ) the estimating difficulty becomes significant. With no experience to serve as a guide, the only solution is to make the best possible guess, using whatever relevant experience available. It seems clear, however, that the more unique an activity is, the less certain the estimate of its duration and, therefore, the more risky the project will be in terms of control. And since a lot of projects (such as research and development) fall into this category, the question naturally arises as to whether some method might be employed to reduce estimating risk. It was in response to this problem that PERT was developed around 1957 as a joint effort between the U.S. Navy and the consulting firm of Booz, Allen, and Hamilton. The concept was originally applied to the Polaris submarine project. Although estimates of activity durations for CPM projects are taken as averages based on history, once they are in place, they are often assumed to be more or less fixed or, to use the colloquial expression, they are “engraved in granite.” The PERT system, however, is based on the recognition that estimates are uncertain, and therefore it makes sense to talk of ranges of durations and the probability that an activity duration will fall into that range, rather than assuming that an activity will be completed in a fixed amount of time.