production of the Prismatic Structure of Basalt. 131 



Origin of Volcanic Heat and Energy " (Phil. Trans. Part 1, 

 1873), from which results he has since calculated partial mean 

 coefficients for each 500°, or thereabouts, within the above 

 limits. The partial mean coefficient for temperatures between 

 900° and 600° F. gives a lineal contraction =0*00000396 per 

 unit of length for 1° F. We may therefore provisionally con- 

 clude that basalt in cooling from the brittle point, or 600° F., 

 may sustain a contraction for each 100 degrees of temperature 

 lost of 0-000396 of a foot taken as unit, or of nearly 0*4 of a 

 millimetre per metre in length. We know not at what amount 

 of tensile strain per square inch of section basalt, within the 

 limits of temperature here in question, will be torn asunder by 

 mechanical tension, nor have we any information as to what is 

 the amount of extension of the material at rupture, though we 

 may be pretty certain that its tensile strength is considerable 

 though probably inferior to that of glass, and its extensibility 

 extremely small though probably greater than that of glass. It 

 is certain, however, that whenever, upon a line as a b (supra) 

 of whatever length, the amount of contraction by cooling 

 through a given range of temperature exceeds the amount of 

 extension upon the same length that would produce rupture, 

 then splitting in one or more places between a and b must 

 occur. It does not seem probable that the ultimate extension 

 of basalt at rupture at any temperature below 600° would reach 

 as much as 2 millims. per metre ; and if so, a reduction of 

 500° Fahr. in temperature would be more than sufficient to 

 originate splitting. It is obvious also that the diameter of the 

 prisms in any given instance depends mainly upon the nume- 

 rical relations of these coefficients of contraction and of extensi- 

 bility. The upper-surface film of the mass of basalt having once 

 commenced to subdivide itself, as illustrated with exaggeration 

 in fig. 2, the splitting proceeds downwards into the mass pari 

 passu with its cooling as exhibited in fig. 3, and, at an after-stage, 



FJC2 



FIG* 3 



T-J 



i I i r 



in fig. 4. But the cooling of the mass no longer takes place 

 from the top surface only as at the commencement ; refrigera- 

 tion now takes place also from the faces (so far as they are opened 



K2 



