398 MR C. RANKEN AND DR W. W. TAYLOR 



below the clock record. In most of our experiments the speed of the paper was so 

 adjusted that 15-20 millimetres corresponded to an interval of one second.* 



As the determinations of the corresponding times of flow for the solvent and 

 the solution were always made on the same day, errors due to variation of the rate of 

 the clock were to a large extent eliminated. 



The following consecutive determinations of the time of flow of a volume of liquid 

 will give some idea of the accuracy attained : — 



(1) Water, 166-95, 166-93, 166-96 seconds. 



(2) Solution of carbamide, 157-77, 157-84, 157"78 seconds. 



(3) Water (30°), 206-50, 206-55, 206-30, 206-42, 206-43 seconds; mean, 206*45 seconds. 



No. (3) is given as an example of the extreme variation observed. In such cases 

 the experimenter is often aware that he has missed the precise moment ; but, as in 

 the above set, exclusion of the lowest, or of the two extremes, has practically no effect 

 on the mean value. 



Before each experiment the viscosity tubes w T ere cleaned with chromic acid mixture 

 and thoroughly washed with distilled water. In order to avoid the use of alcohol and 

 ether, the tubes were dried by connecting them with an electrically -driven Geryk 

 pump, the air being passed through cylinders with calcium chloride, phosphoric 

 anhydride, and cotton wool before being admitted to the tube. If, on repetition, the 

 results were not found to be concordant, it was generally found that the discrepancy 

 disappeared after the tube had been recleaned. 



The degree of accuracy attainable in introducing the same volume of liquid into the 

 viscosity tube was also carefully inquired into. This was necessary, for it seemed 

 probable that the volume of liquid delivered by the same pipette might be different for 

 the solvent and the solution, and for different concentrations of the solution, owing to 

 the differences in viscosity of the liquids. 



In the test experiments the viscosity tube was weighed before and after the addition 

 of the solvent and of the solution, and the weights of liquid were compared with their 

 densities as determined in the usual manner. 



(1) 15° C. H 2 0, (a) 2-1412 gm. ; (b) 2-1416 gm. ; (c) 2-1410 gm. 



(2) 25° C. H 2 0, (a) 2-7215 gm.; (b) 2'7201 gm. 



•05 m. KC1, (a) 2-7247 gm. ; (b) 2-7262 gm. 

 d 2574° = 0-9988, instead of 0-9993. 



(3) 15° C. H 2 0, (a) 2-7239 gm. ; (b) 27222 gm. 



•25 m. KC1, (a) 2-7560 gm. ; (b) 2-7568 gm. 

 d 15°/4° = 1-0114, instead of 1-0112. 



It may be taken that the volume delivered was constant to within 1 in 2000. 



In conclusion, a few actual experiments are appended, which were made on totally 



* [This form of electric chronograph, which is commonly used in physiological laboratories, is very convenient 

 for physico-chemical purposes ; for, if required, it can make a large number of independent experimental records 

 simultaneously, without alteration of the apparatus beyond an additional number of electromagnetic time-markers, 

 which are small and inexpensive. The apparatus was made for us by J. Ednie, mechanic in the Physiology Depart- 

 ment, Edinburgh University.] 



