FLUIDITY AND TEMPERATURE 137 observed values is 0.17 for the 87 substances and based on some 1,OOO duplicate observations. If we omit the alcohols, this difference falls to 0.09 for 70 substances. This is much better than Thorpe and Rodger obtained with Slotte's , since the percentage difference is nearly twice the , viz., 0.15 per cent for 64 substances. But the real test is with substances which give fluidity curves departing widely from the linear type and here Slotte's equation breaks down completely. For this type of substances, the fluidity Eq. (53) with three constants does not reproduce the observed values to the limit of experimental error, but a great improvement can be made by introducing another constant and writing the equation T-Av + C- (54) For example, the mean divergence between the observed and calculated values for the eight substances, which gave the largest percentage difference, was 0.77 per cent with the simpler for- mula; the Eq. (54) with four constants reduces this to only O.O7 per cent which is nearly within the limits of the experi- ixiental error. In the case of water, which gave a mean difference of only 0.17 per cent with the simpler formula, the difference is reduced to 0.01 per cent and similarly in the case of octane it is reduced from 0.16 to 0.02 per cent. For reference, the constants for these substances are given in Table XXXV. XXXV.—THE CONSTANTS IN THE EQUATION T = A

utyric acid .............. 0.23862 43,665.0 433 . 17 200 0.06 JPropionic acid anhydride ..... 0.23619 52,294.0 425 . 82 250 0.07 IBTityl alcohol ...... .......... 0.23605 4,802.0 349.71 40 0.04 Isotoutyl alcohol ............. 0.23700 2,993.7 340.66 30 0.09 -A.csti.ve amyl alcohol 0.24650 2,942.8 346 . 82 30 0.08 Inactive amyl alcohol ........ 0.24101 3,908.7 354.17 35 0.09 H>i methyl ethyl carbinol ...... 0.22988 2,124.0 328.84 30 0.09