316 W. F. SMEETH. 



Perhaps the most obvious characteristic is that the curves con- 

 stituting a perlite never meet otherwise than tangentially, while 

 another striking feature is the fact that the curves are so frequently 

 segmental, the crack ending abruptly in the medium, and even 

 curving round so as to run parallel with an adjacent crack instead 

 of splitting into it. These are points which may be of use in 

 determining whether any particular set of cracks are of a perlitic 

 nature or not. 



Before concluding these remarks I feel tempted to offer a 

 suggestion as to the origin of this unique structure. It has been 

 suggested, and I think generally admitted, that it is the result of 

 contraction of the magma due to cooling. This however does not 

 carry us very far, for everyone is familiar with countless instances 

 of materials cooling, contracting, and splitting up under the strain, 

 without giving rise to a semblance of perlitic structure. Even 

 amongst volcanic glasses — the natural home of perlites — the range 

 of the structure is extremely limited in comparison with the extent 

 of apparently suitable magmas which have been examined. 



In experimenting with Canada balsam, the fact that the struc- 

 ture is easily produced when a ground surface of glass is used, 

 and is only poorly and with difficulty produced on a polished 

 surface suggests at once that the friction of the roughened surface 

 plays an important part. Its effect will be a resistance to the 

 natural contraction of the balsam, giving rise to a state of tension 

 among the molecules. On a smooth surface of glass this tension 

 exists only to a slight extent, and as soon as the polygonal crack- 

 ing takes place, the strain is relieved and nothing further occurs. 

 With the ground surface it is different, and two cases present 

 themselves. First, when the balsam has not been very highly 

 heated (and is therefore not very brittle) and is cooled slowly. In 

 this case the polygonal cracks appear first, and the perlites spring 

 into existence after an appreciable lapse of time. It is evident 

 here that the balsam in the interspaces of the polygonal cracks 

 remains under strain owing to friction, that the strain is towards 

 the centre of each area and increases with the fall of temperature. 



