54 



PROCEEDINGS OP THE AMERICAN ACADEMY 



The theory which I have formed to account for these facts is as follows. 

 We may conceive that the molecule Fig. 4. 



of mica is an ellipsoid, whose ellip- 

 tical section, through the longer 

 axis, can be inscribed in the rhomb 

 of 60° and 120°. Assume now 

 that these molecules have polarity, 

 the rhombic prism would be the 

 normal result of their association, 

 as represented in Fig. 4. We 

 may, however, conceive that three of these molecules may become 

 associated by alternate single poles to form a nucleus, such as repre- 

 sented in Fig. 0. Such a group once formed would be in a condi- 

 tion of great stability, resulting from the 

 concurrent action of the several poles ; 

 and if now each of the molecules devel- 

 ops into a crystal, the result would be a 

 made of a form which is very common in 

 the mineral kiuirdom. If six molecules 

 unite in a similar way to form a nucleus, 

 as in Fig. 6, we should also have stable 

 equilibrium (although less firm than 

 before), and the result of a symmetrical 

 development would be a made such 

 as has been represented in Figs. 2 and 

 3. Whether this more complex arrange- 

 ment is necessary in order to explain the phenomena presented by 



Fig- 5. 



tainty of plates consisting of laminae which verj' nearly fulfil the conditions we 

 liave described. Very small inequalities of thickness renders the effect irregu- 

 lar, and made it at first difficult to discover the law. Our experiments have 

 been a series of approximations ; and, although we may never be able with our 

 rude appliances to compete with nature in the manufacture of uniaxial crystals, 

 yet we have approaclied so near to the perfect result as to be able to point out 

 with confidence one way, at least, by which the effects seen in natural crystals 

 may be produced. We have usually cut the lamiuc-e into regular hexagons, but 

 equilateral triangles might more easily be cut, and would probably give as good 

 results ; for, although the errors of position niiglit not be so well distributed, we 

 have found that a slight variation in the relative position of the lamina injures 

 the result to a far less degree than the least inequality in their thickness. It 

 still remains to make careful quantitative measurements of the effects produced 

 under determinate conditions ; and, as can easily be seen, the subject opens a 

 wide field for mathematical analysis as well as physical investigation. 



