Temperature on the Viscosity of Gases. 221 



given for different methods of computing one series of data. 

 It is perhaps worthy of remark, that the mean of Meyer's 

 results at 100° agrees with the results of my fifth series 

 within 0*2 per cent. Meyer's observations were made chiefly 

 at two temperatures, and thus, as well as on account of the 

 magnitude of the errors of observation, are incapable of indi- 

 cating the small deviation of rj t : r] Q from the equation of the 

 first power of t . 



The observations of Puluj are shown by the lines K L, 

 M N, P, Q R, and S T, of which all but S T are from trans- 

 piration experiments, this being by oscillating plates. From 

 the data which I have used in plotting the lines, Puluj 

 deduces the first power equations : — 



KL, 77=0000179 (1+0-0024 t), Range 180 to 27-0. 



MN, 77 = 0-000179(1 -f 0-0023*), „ 13-6 to 76-7. 



OP, 77 = 0-000181(14-0-0022*), „ 1-1 to 77-4. 



QR, 77 =0-0C0180 (14-0-0021*), „ 15 to 92-7. 



ST, 77 = constant (0-038554-0 00010213 *) „ -3-0 to 25-0. 



From the plots on Plate II., and from others which I have 

 made, it appears to me that the data shown in K L cannot be 

 represented by a straight line, nor are they to be relied upon 

 for giving the rate of change of rj t : r) at all, the curvature 

 of the line K L, which approximately represents the data, 

 being undoubtedly the result of uncorrected constant error. 

 The points of the line MN show an upward convexity of 

 slight amount ; those of the line O P are so irregularly distri- 

 buted as to conceal any slight systematic deviation from the 

 straight line. The points of Q R are also very irregularly 

 distributed, but are unquestionably more nearly represented by 

 the line Q R, which I have drawn approximately through them, 

 than by the straight line SRj, which correspondsto the equation 

 given by Puluj, or by any other straight line. The line Q R 

 should certainly be convex upwards. 



The numerous experiments of Obermayer give almost 

 unquestionable evidence that even for air the coefficient of 

 viscosity increases at a less rapid rate at higher than at lower 

 temperatures. The line a S b c represents the mean of results 

 by capillaries " 11/' " 26," and " D." (I was obliged to 

 omit " I" from consideration, owing to an apparent error. 

 Wien. Ber. Ixxiii. p. 440.) The separate results are fairly 

 accordant. This line agrees quite closely with my fifth series, 

 SY. The results obtained with a brass capillary are indicated 

 by the line d e, which is a continuation also of S b, since the 

 brass capillary agreed substantially with the others at b. The 

 curvature of this line, as well as of a $ b c, is decidedly convex 



