238 Mr. W. Phillips, on the angles of the 



twenty others, of each of which I possess very numerous isolated 

 crystals. Amongst these, the red oxide of copper already men- 

 tioned is one, though of comparatively little importance ; because, 

 as the real value of the angles of the cube, which it occasionally 

 assumes, are known, the rest may be calculated with precision. 



But there is still another circumstance which forcibly shews the 

 great difficulties attending the accurate admeasurement of the angles 

 formed by the meeting of the natural surfaces of crystals, by sub- 

 jecting those surfaces to either goniometer. It is not often found 

 that the best selected afford incidences corresponding with the 

 results obtained by calculation, even where we must be assured that 

 calculation may be relied on ; as is the case in respect of all those 

 substances which assume the cube, the regular octohedron, the 

 rhomboidal dodecahedron, or the hexahedral prism. The angles 

 formed by the meeting of the planes of these solids are known, yet 

 rarely do the best selected crystals accurately afford the admeasure- 

 ments which belong to them in their perfect state. It is not suf- 

 ficient that they approximate ; they should be precise. 



The foregoing are my principal reasons for adopting the belief 

 that many of the determinations of Haiiy in regard to the ad- 

 measurements given by him of the primitive crystals comprehended 

 in the term parallelopiped, and of the numerous varieties of the 

 octohedron, will be found inaccurate. It is probable that the 

 reflecting goniometer will render service to science by detecting 

 some of the inaccuracies incident to the use of the common go- 

 niometer. 



The accuracy of which the reflecting goniometer is capable, 

 requires, for the reasons above assigned, great patience and per- 

 severance in measuring the angles of crystals on their natural 

 planes ; and the chief, if not the only, difficulty attending its use, 



