in the Summer of 1911. 471 



Reference to the curve will show the first abnormality to 

 have been a sudden upward oscillation of the lava in the night 

 of July 18th, with repetitions on August 25th and September 

 13th. These consisted of a phenomenal irruption of gas from 

 below, temporarily raising surface conditions to the super- 

 normal grade of activity and causing a rise — almost momentary 

 in its duration — of some two meters in the level of the lake. 

 Each of these gaseous outbursts was followed by a very con- 

 siderable subsidence of the lava column, as will be seen from 

 the curve, and the phenomenon is, in all probability, purely 

 endo-volcanic, having its cause in the depths of the conduit. 

 If we consider this as widening out into its parent fissure at no 

 great depth it is easy to imagine gas retention in the latter 

 and, from time to time, an escape into the conduit. That an 

 accumulative process is involved seems indicated by an appar- 

 ent periodicity in these outbursts, the interval between the last 

 two being double that of the first. The writer finds, moreover, 

 on consulting his notes, that a repetition of this phenomenon, 

 but on a very small scale and not observed by him, was reported 

 to have occurred in the night of August 6th, when the lake 

 surface was said to have suddenly risen " about one foot," with 

 considerable agitation. At the time of plotting the curve this 

 event was not considered of sufficient magnitude to be intro- 

 duced, but, admitting it as an actual recurrence of the phenom- 

 enon, we have a regular series of these gaseous irruptions 

 with a period of about eighteen days. That of August 25th 

 was coincident with a series of local seismic disturbances, four 

 shocks preceding aud two following the outburst of gas and the 

 entire period of general agitation lasting from 6.30 a. m. to 

 2.35 p. m. The absence of seismic effects at the times of the 

 three remaining oscillations forbids our considering the earth- 

 quake as the cause — temptingly easy on the theory of the con- 

 cussion causing coalescence of the smaller gas vesicles distributed 

 through the lava, by which means, alone, these may acquire 

 effective dynamic properties — and makes it consequent or con- 

 comitant to the gaseous outburst. These rapid oscillations of 

 the lava were instrumental in the sinking of the floating island 

 of 1911, as described in the paper on that subject.* 



The effect of atmospheric pressure variations upon the 

 column of lava is, as a rule, more apparent in its control of the 

 rate of gas flow than in any observable variation in the height 

 of the lava column excepting as this is influenced by the rise 

 and escape of the gas. The case of Stromboli is we'll known, 

 but all the Italian volcanoes — especially when in the condition 

 of actual eruption — show increased activity on a decrease of 

 atmospheric pressure, even the diurnal minima exerting an 

 * This Journal, March, 1913. 



