240 , 62 



time, because the pressure of tlie steam from lielow drives tlrem along as soon as 

 they are formed, hut they are, however, ol suflicienl duration to account for the 

 intermittent character of tlie eruptions. 



Wlien the eruption is on the decline and the evolution of steam decreases, 

 tlie water barricades can remain in the channel a little longer. The spring must 

 have time to gain strength to expel them, but when at last it is able to master 

 them, it will eject them from the channel with great velocity. This will doubtless 

 account for the extraordinarily high water jets which we noticed so often at the 

 end of the eruptions of the two geysers at Grafarbakki. 



During the eruption the great water pressure usually prevailing in the spring 

 cavities becomes almost nil. The accumulated energy of the spring is therefore 

 liberated. But when this is exhausted, and the steam evolution is only sustained by 

 the heal energy continually supplied from beneath, the steam current through the 

 channel is so much weakened that it is not able to keep the passage through the 

 channel free. 



The barriers of water formed as described, in bends and narrow passages of 

 the principal spring channel, can then remain stationary, as long as there is a 

 perceptible stoppage of the steam current through the channel. But as soon as the 

 steam current is stopped, the water will also fill other parts of the channel, from 

 which water has previously been excluded by the steam. The increase of the 

 water pressure then becomes greater than the increase of the steam pressure from 

 below. As a result the water column recedes downwards through tlie channel, 

 enters the cavities of the spring, and fills up the whole of the space formerly 

 occupied by the steam. The steam is unable to resist, on account of the pressure 

 of water being so great and the tension of the vapour becoming less through the 

 decrease in temperature caused by the entrance of colder water from above into 

 the cavities. The eruption is now ended, ^yhich event is marked by the water 

 from tlie basin rushing down into the channel. 



We have noticed that the spring gas, saturated with steam, which is con- 

 tinually supplied to the spring from below, cannot force its way in the form of 

 bubbles through the water in the spring channel, without producing a decrease in 

 the pressure of the water, and thus causing an eruption. In exactly the same way, 

 the eruption is brought to a close through the steam current kept up by the con- 

 stant supply of energy to the spring becoming insufficient to keep the spring free 

 of water stoppages, which block the passage of the steam current through the 

 channel. 



This is, in my opinion, the principal cause of the periodic eruptions of the 

 thermal springs. Of course, very hot water, by suddenly boiling, may also cause 

 an eruption, as many of the geyser models show, but it occurs very seldom in 

 nature. On the whole, it is hardly pi'obably that the spring water can be super- 

 heated to any extent, because it is, in almost every case, impregnated with spring gas. 

 In those cases where hot springs are found with a temperature a little higher than 



