Reciprocity principle is a common theory in electromagnetics and optics. It is also one of the general principles of the radiation transformation theory. However, in many remote sensing studies, this principle cannot be always supported by experimental data, especially when the scaling effect of the remote sensing image pixel is considered. The debate over the issue of whether the reciprocity principle can be used as a fundamental standard in evaluating the effectiveness of remote sensing observations has lasted for years. Using geometrical optics model, Li and Wan proved the existence of the scaling effect in applying the reciprocity principle to a remote sensing image pixel in 1998. In 2002, Snyder challenged Lis proof, and attempted to prove that reciprocity principle is universally valid without any scale limitation. In this paper, we will argue with Snyders theory and point out that Snyders theory had neglected an important condition used in Lis proof and, as a result, drawn the wrong conclusion. Here we will restate Lis condition in his demonstration and offer a further proof to verify that reciprocity principle cannot be applied unconditionally to the study of bi-directional reflectance distribution function (BRDF) of land surfaces in field or satellite remote sensing observation scale.