A comprehensive series of experiments was run on the more significant physical arrangements of the sensitive surfaces of photocells to determine what combinations give the most perfect sampling of light. All tests were run on a duplex recorder described earlier. Two light stations only a few feet apart on the roof of the laboratory were used in these tests. When all conditions were controlled, identical daily records for the 2 stations could be obtained. Opalescent shades were used over the photocells in most of the exps.; except with spherical shades, serious errors of sampling occurred. 6 types of photocell combinations were tested the 2 found to have the least error were a 6-celled unit in which a photocell was fastened to each of the 6 sides of a cube and the whole enclosed beneath a spherical integrating shade of opal glass, and a 1-celled photounit in which 1 photocell was placed at the base of an integrating sphere facing up. There was little significant difference between the sampling of these 2 units but certain other considerations make the 1-celled unit preferable. The determination of visible radiation by exposure of the receiving unit horizontally to the direct rays of the sun, as is done by the weather bureau for total radiation, does not seem suited to physiological studies owing to the impossibility of correcting by the cosine law for lateral light and other effects; also the exposure of photocells to the direct rays of the sun may be very detrimental to them. A suggested method for integrating light records for various foot-candle levels to enable each investigator to choose those levels of significance in his studies is given.