This paper demonstrates, in summary form, some corrections to be made to values of spectral flux as measured with Eppley spectral pyranometers. The results of an investigation into the occurrence of what are believed to be systematic errors in scaled hemispherical thermopile systems, employed in pyranometer design, are presented. This investigation was prompted by an analysis of a large mass of solar spectral measurement material, which indicated an apparent change in spectral pyranometer sensitivity under conditions of continuous operation. The study also included pyranometers of natural- and forced-ventilated design. The results indicate that the corrections so far established ought to be of general application, within practical limits, to the present Eppley spectral pyranometer and, perhaps with some modifications, to other constructions.