1!»18] on Studies on Liquid Films 395 



direct evidence of greater rate of transference in the smaller bubbles ; 

 for when records are taken of a column of unequal segments it is 

 easily seen that the initial inequalities are accentuated as time goes 

 on. Figs. 30 and 31 are reproductions from photographs of two 

 small columns at intervals of nine and seven days respectively ; the 

 volumes of the segments varied from 30 to 100 cm. 



Large l)lack volumes were more difficult to deal with on account 

 of irregular oscillations ; their relatively small mass when black was 

 insufficient to stabilise them against any local convection. For 

 example, in the 200 litre globe already described a column was blown 

 consisting of six segments, the upper five of which had each a 

 capacity of Ij litres, while that of the lowest was 3 litres. In one 

 day it became wholly black with the exception of a thick coloured 

 area that extended up over 30 per cent, of the lowest bubble, the 

 weight of which was sufficient to keep the whole hanging vertically, 

 although some of the junction planes were slightly inclined. By the 

 sixth day the colour had diminished to a zone 5 cm. wide. This 

 reduced load could not keep the column straight, and some of the 

 oscillations that occurred displaced the drop on the lowest bubble as 

 much as 15 cm. from the vertical, while a vertical contraction of 

 15 mm. had accrued, followed in the next two days by a further 

 contraction of Qh mm. The coloured zone was now a disc of 2 cm. 

 diameter, and the column had become so curved that an unusually 

 large oscillation caused the lowest bubble to touch the globe, with 

 the result that only the two uppermost segments remained, the 

 junction plane between them being inclined at 3o' to the horizontal. 



A light glass ring was therefore arranged below to keep the 

 column stationary. It was sealed to a 3 mm. glass rod bent round 

 in a bow, roughly to follow the contour of the globe, and then 

 secured to a vertical glass rod that could slide air-tight in the rubber 

 stopper at the neck. The contraction of the column due to the gas 

 transference, however, went beyond the limits through which the 

 glass rod could be raised. The segments therefore became thinner 

 and more strained, until on the 89th day separation took place 

 between the third and fourth. The glass ring thereupon relaxed 

 a distance of 8 mm., the amount by which the glass bow had 

 been strained by the pull of the black column. 



In some cases the various segments were measured daily by a 

 cathetometer and horizontal sliding telescope. If I) is the diameter 

 of a bubble segment at the equator, and d the diameter of the plane 

 circular ends,' while h is the height, then the volume is given by 



y = 1'0472/i (0-4 D^ + i)"2 J)d + O'lbd^). 



When the segment is unstrained, d ^ D - 0*268A, so that 



y = l-0472A(0-75 D- - O'lSO J)h + O'OU-); 



only D and h need therefore be measured. 



