238 60 



staut in all the hot springs, whether they be periodic or in a constant state of 

 ebullition. For I am nol aware of any reason why this supply should be periodic, 

 and this assumption is rendered more plausible by the fact that both periodic and 

 constantly boiling springs are often situated side by side in the same group of 

 springs, as, for instance, at Grafarbakki. It is therefore only in the shape and 

 position of the channel that one can expect to find the factors which cause the 

 periodic springs, — as distinguished from the constantly ebullient springs, — to 

 eject the liot water and spring gas periodically. 



The fact that no gas bubbles rise through the water in the basin immediately 

 after an eruption proves that the spring gas, — which appears to be an inseparable 

 attribute of every Icelandic hot spring, is blocked up down in the spring channel. 

 I must therefore assume, with Mackenzie, that there are cavities down in the 

 earth, in which the spring gas is collected. And it is highly probable that the 

 spring water, by dissolving the rocks, has formed many such cavities down in the 

 earth, inasmuch as the spring water usually contains great quantities of dissolved 

 substances. 



Although I think it is most probable that the majority of geysers have several 

 cavities connected with the spring channel, I will, in order to simplify the following 

 remarks, restrict myself chiefly to the case of only one cavity, as shown in Fig. 14. 

 It is evident that the same argument will hold good when the spring has many 

 cavities, although the whole subject then becomes more complicated. The water 

 in the bend of the channel B cuts off from the atmosphere the spring gas accu- 

 mulating in A. But as fresh spring gas is constantly coming from below, the 

 gas accumulated in A is constantly expanding. Again, the loss of heat in the 

 spring being comparatively small while the channel is stopped up, and the supply 

 of heat to A being constant, the temperature in it must consequently increase. 

 This causes an increase in the volume of the gas in A, on account of the expansion 

 of the gas itself, and the augmentation of the steam pressure through the increasing 

 temperature. This increase in the gas volume in A is at first counteracted by the 

 increasing pressure produced by the rising of the water in the channel B. At last 

 the volume of the gas in A increases so much that bubbles of spring gas, saturated 

 with vapour, begin to force their way through the channel B. The volume of the 

 first bubbles is, however, very much diminished on the way up, on account of the 

 condensation of a great deal of the vapour through contact with the colder water 

 in the channel. A considerable difference is possible in the temperature of the 

 water in different parts of the channel, especially when the channel is so narrow 

 that the water cannot circulate freely enough to distribute evenly the loss of heat 

 in the spring water which takes place in the upper part of the channel, through 

 the lower temperature of the surroundings. The condensation of the vapours from 

 the gas bubbles increases the temperature of the water, so that subsequent bubbles 

 are not so much diminished. 



In ordinary circumstances a state of equilibrium will eventually ensue, in 



