The resistance ratio method is the most frequent technique used to determine the extent of interstitial loading of hydrogen or deuterium atoms into palladium electrodes, or extended structures used in electrolytic or gas phase cold fusion experiments. Specifically, advantage is taken of an empirical relationship between the measured resistance, R, normalized to that of the same body at the same temperature in the absence of hydrogen isotope, R°, hence R/R°, and the atomic fraction occupancy of octahedral interstitials, x = H/Pd or D/Pd. This method was first suggested and employed in cold fusion studies by the present authors [1], and received immediate and widespread acceptance because of the ease with which this experimental technique could be used to make insitu, real-time measurements of a parameter, D/Pd, anticipated [2] or hypothesized [3][4][5][6] at that time to relate to cold fusion heat excess or nuclear production.We take up this topic again 15 years later in an attempt to clear up some errors and misunderstandings regarding the resistance ratio method and its application in cold fusion studies. The relationship between R/R° and x is empirical. That is, calibrations are only as good as the experiments that support the shape of the curve and cannot be used outside the range [P, T, x] in which data are taken. The original calibration [1] (unaccountably and erroneously immortalized as the "famous Baranowski curve") involved an extrapolation of known data into the region of cold fusion interest in the D-Pd system, at x>0.6.Present theory and results focus new attention on the very high loading region as x approaches or even exceeds unity, where double occupation of octahedral sites, tetrahedral site occupancy, new phase formation or new electrical states, may be relevant to the underlying physical process of excess heat and nuclear production. Rather than simply using the resistance ratio as a qualitative tool to determine whether an electrode is better or lesser loaded, it is now important to obtain accurate quantitative information for x close to unity. With further experimentation and analysis of published data it is apparent that the curve originally published in 1990 is in error in the high loading condition. This paper describes how this empirical fit has been improved over the years for both H/Pd and D/Pd by employing new data, new analysis of old data, new experimental methods and results.