234 ROBERT B. SOSMAN 



that can be controlled and measured. Such experiments will 

 require the melting and handling of larger quantities than it has 

 been customary to use for laboratory experiments, but the diffi- 

 culties ought not to prove serious. Even in the absence of such 

 experimental data, much can be learned from a careful field exami- 

 nation of prismatic and columnar structures. 



