438 



The resulting bubbles were photographed by cine-camera (l6 frames/second nominal speed) 

 distances being dettrmined by comparison with the edges of the panes of glass (2i feet), and the mesh 

 of the protective wire-netting inside the tank (3 inches). The time-scale was established by 

 photographing dials driven Oy a phonic motor controlled Oy a tuning fork. 



Description of Bubbles , 



The out-of-balance couple on the bucket had been adjusted in preliminary experiments near the 

 surface of the water, so that all tne air was "poured out" simultaneously. If the couple is too 

 small, the air escapes in the form of small bubbles, while if it is too large the bucket goes beyond 

 the vertical before all the air can escape. when the weight was properly adjusted, the air "poured out" 

 caused a consideraole upheaval and "whitening" of the water, similar to the "plume" from a deep 

 explosion. When the bucket *i«s 10 foet or more below the water surface, it was found that the air 

 came up as about u - lO large bubbles and a large number of smaller ones. The disturbance produced 

 at the surface was very slight. It was concluded that the original large bubble had broken up. 

 When the experiments proper were begun in the glass-fronted tank it was found that the large bubble 

 broke up almost immediately into a small number of large bubbles and a multitude of very small ones. 

 All the large bubbles were of the characteristic "mushroom" shape described by Taylor and Davies (1) 

 and rose at speeds af tne order of 2 feet per second. Some were observed to break up as they floated 

 upwards and it was also thought that some were colliding and coalescing, but the latter impression 

 provided, en examination of the films, to be a mistake. Twobubbles sometimes settled down into an 

 apparently stable configuration, one below the other, but with axes of symmetry offset, and rose 

 together. No larger complexes were seen. 



The bubbles produced by the burning cordite were very similar tc the air-bubbles, and the 

 general sequence of events seemed to be very much the same. It did not appear to make any difference 

 to the general nature of the p.^'enomena whether the cordite was fired in the easily breakable pill-box 

 or the strong cartridge-case. Indeed, one of the largest bubbles obtained in the trial was obtained 

 in the latter circumstances, which one wculd think were the most unfavourable. Some close-up 

 photographs were taken in the smaller tank at Glen Fruin in order to get the earlier stages of bubble 

 formation in mere detail, but the earlier stages were cOscured by what appeared to be a cloud of small 

 bubbles. A few photographs v»ere taken from above of a bubble "breaking surface" but exhibited no 

 feature of interest. The surface phenomena were confined to slight ripples (Figure l(b)). 



Two features noticed by Taylor and Davies (l) were apparent in the records. First, the bubbles 

 are often "lop-sided," i.e. not perfectly symmetrical about an axis, the "lopsidedness", when it exists, 

 often persisting throughout the life of the bubble; secondly, in many of the records, there is an 

 indication of a wake lying approximately in the sphere of which the bubble is a cap, similar to the 

 effect shown by Taylor and Davies (l) in Figure 1 of their report. In our case the existence of a 

 wake is suggested by clouds of small bubbles following each large one (Figure 1(a)). 



Analysis of the Records . 



The trajectories of 31 bubbles of various sizes were plotted. within the errors of the 

 experiment the velocities were all constant in time, in spite of the fact that the depth of water was 

 33 feet so that the bubbles must have expanded to twice their volume during 'the rise. A few of the 

 trajectories showed indications of a slight oscillation (too stbU to measure) aoout a mean velocity, 

 but there was no definite upward or downward curvature. The images of the bubbles were somewhat 

 blurred, but in some cases it was possible to estimate the radius of curvature of the upper cap of 

 the bubble, in order to check up the relation u = 7,1/43 found by Taylor and Davies (l) for their 

 small bubbles. The blurring of cur images may be due to the condensation mist on the glass panes of 

 the tank, which was difficult to remove completely. 



The radius of curvature was measured by projecting a magnified image of the bubble on to 

 squared paper and estimating the radius frcjii the length of the chord and the distance from the chord 

 tc the top of the bubble. owing to the blurring it was not possible to say how nearly the bubbles 

 are to parts of spheres, but it appeared that the departure was not great, and that the estimates of 

 the radii should be accurate to 20J. The radius was measured at several different points en each 



trajectory 



