Slocombe, C. S. (1927). The Spearman prophecy formula. Journal of Educational Psychology, 18(2), 125-126.
http://dx.doi.org/10.1037/h0071069
The over-prediction occurring in the use of the Spearman prophecy formula in several investigations is probably due to the presence of systematic changes in the repetition of test forms, i.e., to practice. When sampling coefficients we must use either random sampling or first and last, second and next to the last, etc., and not first and second, third and fourth, etc. (PsycINFO Database Record (c) 2016 APA, all rights reserved)