Theory of Explosions. 167 



the flask would have been if it had not disappeared. There 

 were so many sources of misinterpretation to be feared in this 

 method that I did not continue it, but noted the peculiar 

 rolling, and the dead-beat motion, of the dye as it was shot out. 



Unfortunately, just as I had reached the conclusion of 

 these observations, one of the windows gave way, and I 

 decided to board them both up, and take some observations 

 with an impact gauge or gauges, which I had meanwhile con- 

 structed. Each gauge consisted of a short hollow cylinder of 

 brass, about 3 inches in diameter and an inch deep. One 

 end was made of stout brass, and the other had a sheet of 

 india-rubber stretched tambourine-fashion over it. A brass 

 tube, \ inch in diameter and 18 inches long, w T as fastened 

 into the cylinder in such a way that when the plane of the 

 india-rubber was vertical, the tube was also vertical. Into 

 this brass tube there projected and was cemented a tube of 

 glass projecting upwards about a meter, and having an 

 internal diameter of about j 1 ^ in. Two such gauges were 

 made, and carefully fastened into, and backed by, wooden 

 supports, on which were carried the graduated scales by whose 

 aid any rise or fall of liquid in the glass tube could be mea- 

 sured. These supports were then firmly fastened to the two 

 adjacent sides of the tank. The centres of the india-rubber 

 faces were as nearly as possible in the same horizontal plane, 

 and about 18 inches below water-level when the tank was 

 full. The gauges themselves were filled with water (coloured 

 with rosaniline) to such a point that the coloured liquid stood 

 about an inch above water-level when the tank was full. 

 Precautions were of course taken to prevent any air being 

 entrapped in the gauges. The explosions were caused en- 

 tirely by fulminate of mercury enclosed in glass bulbs of 

 about \ inch in diameter, and blown as carefully as I was 

 able. The covered wires projected their bare ends pretty 

 accurately to the middle of the bulbs, and these latter were 

 rendered water-tight by passing round the joints a little 

 melted paraffin. 



A wooden rod laid across the top of the tank had an iron 

 rod projecting from it vertically downwards to a depth of 

 about 16 inches. The wires carrying the charge were tied to 

 this rod, and allowed to project beyond its end sufficiently far 

 to place the bulb in the horizontal plane of the centre of the 

 two gauges. 



By means of marks on the sides of the tank and of the 

 wooden rod, the latter could easily be adjusted so as to place 

 the bulb at a point equidistant from the two gauges. This 

 distance was varied in different experiments by shifting the 



N2 



