224 ROBERT B. SOS MAN 



It is of interest to note that this convection-crystallization 

 hypothesis explains the original observation of Watt on the forma- 

 tion of columns in a cooling artificially melted basalt mass (see 

 p. 2 1 5) . He accounted for his columns on the assumption that they 

 were produced by the mutual interference of radially-growing crystal 

 bundles, uniformly spaced in a horizontal plane. Why the crystal- 

 lization centers should be uniformly spaced he was unable to say. 

 The existence of convection prisms in the still liquid basalt provides 

 the missing link in the series of phenomena. Crystallization may 

 have begun at the axes of the convection prisms where a few early 

 separating crystals had collected, or at the corners as observed by 

 Benard; in either case the crystallization centers would be uni- 

 formly spaced horizontally. 1 



If liquid convection is really the cause of all or any of the familiar 

 naturally occurring basaltic columns, then it is important to know 

 what criteria will help to decide the question in a given case. 

 Furthermore, a systematic examination of natural columns will 

 throw light on their history, whatever may be their mode of origin. 

 What are the important characteristics of a given occurrence which 

 should be observed in the field ? 



CHARACTERISTICS OF CONTRACTION AND CONVECTION PRISMS 



i. Attitude. — The original attitude of columns formed by con- 

 vection should be vertical, or very nearly so. Contraction columns, 

 on the other hand, are usually perpendicular to a cooling surface; 

 irregular conditions of cooling, furthermore, may cause them to 

 curve in a great variety of ways. 



2. Dimensions. — Convection columns should be much wider, in 

 proportion to their length, than contraction columns, which are 

 commonly very long and narrow. The columns at Murols described 

 by Dauzere are 1.5 to 2 m. wide and 5 to 10 m. high; those 

 measured by O'Reilly in the Giant's Causeway are 0.4 to o. 5 m. in 

 width; the Causeway columns vary from 3 to 25 m. in total height. 

 Scrope describes columns near La Queuille as much as 5 m. in 

 diameter, and 10 m. or less in height. 2 The common contraction 



1 See Longchambon, Bull. Soc. Geol. France, XIII (1913), 33-38. 

 - Volcanoes of Central Frame, London, 1S58, p. 136. 



