ON SELF-ACTING INTERMITTENT SIPHONS, ETC. 225 



water in the vessel rises above the top of the bend of the siphon, and it 

 will be readily seen that if the siphon is of any size, this will require a 

 large accession of water in the tank, so that the siphon will not work 

 except in cases where there is a large flow of water. 



This difficulty can, to a considerable extent, be overcome by dipping 

 the outer leg of the siphon in water, as shown in fig. 2. The water 

 which runs over the bend of the siphon will then drag a certain quantity 

 of air with it, and drive this air out at the lower mouth of the siphon, 

 and as the air cannot return in consequence of this mouth being sealed, 

 the air in the outer leg: is gradually reduced in tension below the 

 atmospheric pressure. Whether this partial exhaustion of the air in the 

 outer leg is sufficient to start the siphon, depends on the quantity of water 

 that runs over the siphon, but the quantity required will be much less 

 than if the outer end were open, and it will not be necessary for the water 

 in the vessel to rise above the top of the bend of the siphon. 



Although the expedient of dipping the outer leg of the siphon in 

 water greatly induces the quantity necessary to start the siphon, the re- 

 quired quantity is still very considerable if the siphon is of any size, and 

 further expedients have therefore been adopted to reduce this quantity. 

 One of the simplest of these expedients is to have two siphons of different 

 sizes connected together by a tube at the crown, and so arranged that the 

 water runs through the smaller siphon first. The outer ends of both 

 siphons are dipped in water, the smaller siphon then starts with a com- 

 paratively small quantity, and afterwards by means of the connecting 

 tube exhausts the air from the larger siphon, and brings it also into 

 action. This method was adopted by Professor James Thomson, F.R.S., 

 in 1860, for his jet pump, and it was also carried out on a large scale in 

 France in 1867, at the Reservoir de Mettersheim. In this latter case, 

 there are two siphons of about 28 inches in diameter, each of which is 

 put in action by a smaller siphon of 6 inches in diameter. 



This expedient, however, and several others which have been adopted, 

 leave much to be desired, as they are to a certain extent complicated, 

 and yet do not sufficiently reduce the quantity required for starting the 

 siphon to enable it to be used in many cases. The method which I am 

 now about to describe is both simpler and much more effective. 



In an extensive series of experiments which I tried some years ago 

 on siphons, with their outer legs dipped in water, I was much puzzled by 

 finding that the quantity of water necessary to put a siphon of given size 

 in action varied in the most unaccountable way at different times. The 

 only difference that could be perceived between the cases in which the 

 siphon started and those in which it did not start was, that in the former 

 case air-bubbles escaped freely at the mouth of the siphon, whereas in 

 the latter case, under apparently the same conditions, very few bubbles 

 came out. At last the idea suggested itself to me of making a portion of 

 the siphon in glass, so as to see what was going on inside the pipe, when 

 the cause of the irregularity was at once discovered. Sometimes the 

 water which ran over the bend adhered closely to the sides of the pipe; at 

 other times a portion of it would fall more or less clear of the sides. When 

 the water adhered to the sides it produced very little effect in displacing 

 the air, so that only a small quantity of air was driven through the water 

 at the mouth of the siphon. When, on the other hand, the water fell clear 

 of the sides, it produced a great effect iu displacing the air, and large 

 bubbles of air at once escaped from the mouth of the siphon. 

 1879. Q * 



