18 PROCEEDINGS OF THE AMERICAN ACADEMY 



Effect of Pressure. 



In Series IV. for air, the total driving pressure used, p^ — p^, was 

 varied within considerable ranges, to test whether the value obtained 

 of r).2 : »;|, or of K, was thereby affected except through errors of obser- 

 vation. In 3, 4, and 5, p.^ — Pz^=^ 248 mm., and rj.^-.rf^ = 1.2G7 ; in 7 

 and 8, under the same conditions otherwise, p^ — Ps = 376 mm., and 

 rj2: r}^^ =: 1.266, showing no effect due to increase of the pressure by 

 one half. Measurements 9 to 13, made with p^ — p^ ranging from 

 258 to 485 mm., give values of K with an average deviation of about 

 0.2 per cent. Measurements 21 to 48 give values of 772 : rj^ about one 

 per cent lower than those obtained in the remainder of the series, 

 made with greater values of 2^1 — Ps j ^^^ the precision of these 

 measurements is much less than of the later ones. The value of K 

 found from 50 to 54 and 60 to 66, with p^ — Ps=^ ^30 mm. about, is 

 within 0.5 per cent of that found in 9 to 1 2, under p^ — p.. = 258 mm., 

 and agrees precisely with 12, under ^^ — Ps^^ ^61 mm. And, finally, 

 the mean of 97 and 98 differs from 99 by less than 0.25 per cent, 

 though the pressures p^ — p^ are respectively 540 and 250 ram. 



In the second series for carbonic acid, the effect of pressure was 

 somewhat tested in measurements 26 to 38, where p^ — p^ varied from 

 370 to 111 mm. without producing any traceable effect in K. This 

 is the best test made, owing to the greater precision of the measure- 

 ments. Also in measurements 93 to 98 of the same series the 

 pressure was changed from 105 to 207 without materially affecting 

 the results. 



In the first series for air, the average value of p^ — pg was about 

 740 mm., and the mean value of 772 = ni at h = 100° is 1.270, while that 

 of the fourth series, under p,^ — p. = from 248 to 540, is about 1.267, 

 and that of the fifth series, under p^ — p2 = 22o mm., is 1.272, thus 

 showing no effect traceable to the difference of the driving pressures 

 used. 



Deduction from Results. 



The results of my measurements seem to show conclusively, that 

 the variation of the viscosity with the temperature of the gas, in the 

 case at least of dry carbonic acid and of dry air freed from carbonic 

 acid, which may be taken as typical gases, is not proportionate either 

 to the square root or to any numerical power of the absolute tempera- 

 ture reckoned from —273° C They point thus to the inference that 

 all hypotheses yet advanced to account for the variation of the viscosity 



