206 



Prof. Silas W. Holman on the Effect of 



action, at the entrance to or exit from the tube, would be in 

 part eliminated in this method, since the measurements give 

 the ratios of the resistance in two tubes under constant and 

 nearly identical conditions. No complete discussion of these 

 effects has been yet given, and their experimental elimination 

 certainly seems more easy than their mathematical treatment. 

 It seems quite possible that they are still sensible sources of 

 deviation from the assumed law of transpiration ; but as the 

 magnitude of the disturbance is small, and cannot be widely 

 different in the two tubes used in this apparatus, and since 

 the determination of the constants of the tubes dependent on 

 their diameters, lengths, &c. are experimentally made by the 

 same process and under conditions identical with those of the 

 subsequent work, the resulting error must be small. It should 

 be noted also, that the debated question as to the slip of the 

 gas over the inner surface of the glass tube has an influence 

 in my results only to the extent by which this slip is affected 

 by the temperature. 



The freedom of the method from constant error is indicated 

 by the close accordance of the results on dry air as obtained 

 by the first, fourth, and fifth series. The first series was 

 made with an apparatus * totally different from that afterwards 

 used in all details of disposition of parts and of instruments 

 used, except that the tube No. I. of that series was the same that 

 was afterwards cut into three pieces, of which two nearly 

 equal ones served as the capillaries in the fourth and fifth 

 series. Observations with other tubes in the preliminary 

 measurements, however, checked satisfactorily with these. 

 The instruments and arrangements for measuring the pres- 

 sures in the fourth series were wholly different from those 

 used in the fifth series. The results at 100° for the three 

 series are given in the following Table, in which the two 

 last values are taken from the equation deduced for the 

 respective series : — 



Series. 



Vo' 



d. 



I 



1-270 



1-2672 



1-2717 



-0-0017 

 -0-0045 

 



IV 



V 







* See Phil. Mag. for February 1877, p. 84. 



