1902.] on the Constitution of Crystals. 137 



There has been a great want of correlated work of this kind, having 

 for its definite object the elucidation of the relationship between the 

 chemical composition of a crystallised substance and the peculiar type 

 of symmetry which it exhibits. 



The salts which have, up to the present, been studied, are the sul- 

 phates and selenates of the alkali metals potassium, rubidium and 

 caesium, which crystallise in the rhombic system of symmetry ; and 

 thirty members of the well-known monoclinic series of double sul- 

 phates and selenates, containing one of the three alkali-metals just 

 mentioned, another metal of the magnesium or iron type, and six 

 molecules of water of crystallisation. 



The research had two main objects. First, to discover the effect 

 of replacing one alkali-metal by another ; and second, the effect of 

 replacing the lighter element, sulphur, contained in the acid radicle 

 of the salt, by the heavier element, selenium. The three metals, 

 potassium, rubidium and caesium, belong strictly to the same family 

 of chemical elements, according to the now famous periodic classifi- 

 cation of Newlands and Mendeleeff. They are the most electro- 

 positive metals known, and the weights of their atoms are related in 

 a most interesting manner, that of rubidium being the mean of the 

 atomic weights of potassium and cassium. It was, therefore, to be 

 expected that any difference of form or properties brought about by 

 the replacement of any one of these metals by another would be as 

 great as could ever be produced by a change of this kind, and it 

 might be hoped would be adequately great to enable a true idea of its 

 character to be obtained. The very fact that no differences in the 

 angles of the crystals of so-called isomorphous salts had, up to the 

 commencement of this work, been detected with certainty, shows how 

 very small, at most, such differences must be. It will also be evident 

 that for such a research only the most perfectly formed and homo- 

 geneous of crystals must be employed. It is a primary essential that 

 the faces of the crystals shall yield perfect images of the signal-slit 

 of the goniometer, and in order that this may be so they must be 

 absolutely plane surfaces. Indeed, if one may be forgiven a slight 

 equivocation in the spelling of a word, the ladies present may be 

 interested to hear that the beauty of crystals lies in the planeness of 

 their faces. 



The mode of measuring the angles between crystal faces on the 

 goniometer, and at the same time the difficulty offered by imperfect 

 faces, may be illustrated with the aid of this large crystal of quartz. 

 It is mounted upon one of the actual goniometers employed in the 

 work, but the size of the crystal is enormously greater than those 

 actually used in the research, which rarely exceeded the size of a 

 pin's head. Such small crystals are much more free from distortion 

 than larger ones. 



You now see on the screen the image of the goniometer signal- 

 slit reflected from one of the faces of the quartz crystal. This image 



