WRIGHT AND HOSTETTERI CRYSTAL GROWTH 413 



paratus for applying the load, F a weakly magnifying objective 

 lens, G a quartz compensator, H a positive eyepiece, and M the 

 analyzing nicol. 



Preliminary experiments were carried out on alum crystals, 

 but for accurate measurements these crystals proved to be un- 

 satisfactory because of the crystal habit, which is octahedral 

 and does not allow the direct passage of the light through the 

 crystal at the edges, since its interfacial angles are not 90°. This 

 defect was remedied in part by grinding parallel surfaces on the 

 crystal normal to an octahedral plane and protecting these sur- 

 faces by cover 'glass slips; this procedure, however, was satis- 

 factory for quantitative measurements. The observations so 

 far as carried proved that the octahedral layers deposited on the 

 crystal under load are in the same state of strain as the original 

 crystal. In order to obviate the effects of prismatic refraction 

 the refractive index of the solution was made equal to that of 

 the alum by the addition of glycerine, but out of this supersat- 

 urated solution crystals did not grow satisfactorily. 



After unsuccessful experiments to grow large sharp unmodi- 

 fied cubes of sodium chloride, and also of potassium chloride, 

 satisfactory crystals of sodium chlorate were obtained which 

 were sharp cubes without truncations and well suited for the 

 purpose. These crystals were not, however, entirely free from 

 local strain but showed between crossed nicols faint interference 

 colors of first order gray. The interference colors moreover do 

 not change rapidly with change in load, and, to render the 

 changes in path difference more evident, a sensitive tint plate 

 was employed in qualitative observations. 



Experiment 1. A crystal (2 by 3 by 4 mm.) was subjected to 

 a load of about 20 kgm. per square centimeter. The load was 

 not, however, uniformly applied, as was shown by the changes 

 in interference colors from point to point; the unequal loading re- 

 sulted chiefly from the bending of the metal disk (D, fig. 1) 

 which was placed between the crystal and the rod (C, fig. 1) ; 

 also in part from slight irregularities in the surface of the crystal. 

 The interference colors ranged in general from first order gray 

 to white, and were rendered more sensitive to slight changes in 



