A THEOKY OF BULBOUS BUDDING 651 



tion causes the accumulation of vesicles in greater numbers near the 

 upper surfaces of the pillows — a condition that has been recorded by 

 several observers. 



Barrow^^^ has compared vesicular and amygdaloidal pillows to the slag 

 blocks that were used in the construction of the great Tees Breakwater. 

 The slag flowed like water and gave of! no gas as it was run into the box- 

 shaped trucks in which it was molded; but when broken by the waves 

 these blocks were found to be very vesicular, especially at the center, and 

 the vesicles were arranged roughly parallel to the outer surfaces. It was 

 concluded that the gases were held in solution and finally released only 

 on the crystallization of the interior. 



9. HOLLOW PILLOWS 



Should the flow of the interior liquid fail from any cause, such as the 

 freezing of an intervening passage or the opening of freer channels in 

 another direction, a growing pillow may be drained on cracking and thus 

 become wholly or partially hollow. However, most hollow pillows seem 

 to owe their cavernous condition to either an excessive amount of con- 

 tained gases, giving rise to a central steam cavity of large size, or to the 

 breaking down of a spongy core by weathering. This latter process may 

 even precede the breaking away of the outer shell. 



10. EXTENT OF FLOW 



The budding of the front and lateral margins of a nearly stagnant flow 

 and the successive budding of these spheroidal masses to form similar 

 ones, after the manner of some species of bulbous cactus, may lead even- 

 tually to the covering of a considerable area, the extent finally attained 

 depending chiefiy on the length of time during which a moderate supply 

 of sufficiently liquid lava continues. On fiats and gentle slopes the lateral 

 extent of a small flow of this nature will necessarily be small ; but with a 

 long-continued supply of lava the formation of successive layers of pil- 

 lows, one on another, will gradually build up a slope like an aggrading 

 river, and with increasing gradient the range of flow will be increased. 

 Thus both the depth and lateral extent of a pillow lava are contingent 

 on the length of time during which a suitable supply is maintained. 



The relative amounts of spheroidal and massive lava in any flow will 

 depend on many physical conditions, among which the composition of 

 the lava, the rate of extrusion, the initial temperature, and the tempera- 

 ture range within which the lava remains a mobile liquid are probably of 



>^ G. Barrow : Quar. Jour. Geol. Soc. London, vol. 64, 1908, p. 271. 



