There is a need to study the likely effects of global warming on crops with experimental treatments as representative as possible of future environmental conditions. One approach that shows much promise is the use of hexagonal arrays of infrared heaters to warm canopies of vegetation. This approach is currently being used in several experiments on low-stature plants (< 1 m) in 3-m-diameter plots (and smaller). However, in order to obtain more sampling material and more representative sampling possibly with sub-plots of other treatments, it is necessary to greatly increase plot size. By nesting hexagonal arrays in a honeycomb pattern, we show that theoretically excellent uniformity of the downwelling thermal radiation (and consequent warming) can be achieved across the plots. Switching from heaters with a characteristic dimension of 60 mm, such as currently being used, to larger heaters with a characteristic dimension of 1016 mm would increase the theoretical heater radiative efficiency (percentage of input electrical energy emitted as thermal radiation) from about 69% to 80% at a wind speed of 4 m s-1. More importantly, the theoretical geometric efficiency (percentage of thermal radiation that falls within the useable plot area) would increase from 37% to about 66% as plot size diameter increases from 3 to 20 m and beyond. The overall efficiency is the product of the radiative and geometric efficiencies, which therefore could change from about 26 to 49% at the 4 m s-1 wind speed as one changes from a single hexagon of the smaller heaters over a 3-m-diameter plot to a honeycomb of the larger heaters over 20-m-diameter and larger plots. Assuming, for example, that 4 degrees C is the desired degree of warming, that electricity costs $0.1 per kW-hr, and the vegetation is dormant 7 months of the year and actively transpiring for 5 months of the year, under Bondeville, Illinois conditions, annual power costs would be on the order of $6,700 for a 3-m-diameter plot and $124,000 for a 20-m-diameter plot. However, there is an economy of scale such that the costs would be about $900 and $400 per m2 for the 3- and 20-m plots, respectively.