Blue stragglers in globular clusters are abnormally massive stars that should have evolved off the stellar main sequence long ago. There are two known processes that can create these objects: direct stellar collisions and binary evolution. However, the relative importance of these processes has remained unclear. In particular, the total number of blue stragglers found in a given cluster does not seem to correlate with the predicted collision rate, providing indirect support for the binary-evolution model. Yet the radial distributions of blue stragglers in many clusters are bimodal, with a dominant central peak: this has been interpreted as an indication that collisions do dominate blue straggler production, at least in the high-density cluster cores. Here we report that there is a clear, but sublinear, correlation between the number of blue stragglers found in a cluster core and the total stellar mass contained within it. From this we conclude that most blue stragglers, even those found in cluster cores, come from binary systems. The parent binaries, however, may themselves have been affected by dynamical encounters. This may be the key to reconciling all of the seemingly conflicting results found to date.

Accepted to Nature, 9 pages, 3 figures