The Infrasound Experts Group of the Geneva Conference on Disarmament Ad Hoc Committee on a Nuclear Test Ban has recommended an infrasound array design consisting of four elements, with three elements forming an equilateral triangle and the fourth at the center of the triangle. The Experts recommended that the sides of the triangle be in the range 1 to 3 km. In this report, in an attempt to place constraints on the array aperture, we evaluate the beamforming azimuthal estimation error of such arrays and compare it to historical data on observed azimuth residuals. The analysis for perfectly correlated signals shows that for beam signal-to-noise » 1 the rms error is proportional to wavelength and inversely proportional to array aperture, beam amplitude signal-to-noise, and the square root of the time-bandwidth product. For beam signal-to-noise « 1 the relations are the same except that the rms error is inversely proportional to the square of the amplitude signal-to-noise.