``Sub-arrays'' are nonstandard configurations obtained by breaking the VLA into smaller arrays that are devoted to different observing programs at the same time. The use of sub-arrays is generally not as efficient as time-sharing the entire VLA. The number of interferometer pairs in a sub-array is N(N-1)/2 where N is the number of antennas in the sub-array. Sub-arrays with 13 and 14 antennas therefore have 78 and 91 interferometers respectively, whereas a 27-antenna standard configuration has 351. An hour of observing in which two such sub-arrays each perform different tasks therefore produces 169 interferometer-hours of data. In contrast, two half-hours of observing, with the full VLA devoted to each task in turn, produce 351 interferometer-hours of data. Dedicating two roughly equal sub-arrays to different tasks thus reduces the amount of information gathered by a factor of about two, compared with time-sharing the whole VLA between the two tasks. This loss of information manifests itself as poorer sensitivity and u-v sampling in the sub-array data. The use of sub-arrays is therefore generally undesirable unless you need strictly simultaneous observations of strong sources at several frequencies (e.g., to measure instantaneous spectra of rapid variables) or to study many compact sources quickly with only modest demands on sensitivity and dynamic range (e.g., for astrometry of strong sources).