THE ACTION OF CYLINDRICAL OBJECTS. 



351 



The transverse section of the cylinder gives us, therefore, the 

 means for deciding whether the axial directions of the ellipses of 

 elasticity coincide with the radius and the tangent, or inter- 

 sect obliquely, and in the former case informs us which is the 

 greater the tangential or the radial axis. But what have we 

 gained by this ? Granted that the two axes really are parallel 

 to the radius and to the tangent, then it must further be asked 

 whether perhaps one of them is at the same time the axis 

 of the ellipsoid, or whether we are dealing with a diametral 

 section ? This question is not always easily answered. The 

 above-mentioned test, by means of rotation upon the two axes, 

 can indeed guide us only in cases where the section is not too 

 thick in proportion to the diameter of the cylinder. With 

 thicker pieces this process is not applicable, for when they 

 are appreciably oblique they no longer act as sections. In order 

 to obtain further data, we shall, therefore, in the majority of 

 cases, have to resort to longitudinal sections, and, where we 

 cannot obtain these, longitudinal views. 



In the first place, as regards longitudinal sections, it is evident 

 that a middle lamella B B (Fig. 202), if it lies flat upon the object- 

 stage, determines the axial position of the ellipse of elasticity in 

 the diametral plane of the cylinder. For since the marginal por- 

 tions of such a lamella act approximately as a plate of crystal, 

 we shall immediately learn whether the two axes of the longitu- 

 dinal and latitudinal directions are parallel, or whether they cut 

 these directions obliquely. We may pro- 

 visionally disregard this latter case wholly* 

 since an intersection of this kind never 

 occurs, so far as we can judge from our 

 observations. 



We may hence assume that the ellipses 

 of elasticity effective in the latitudinal and 

 longitudinal sections have their radial axis 

 in common. The plane through the other 

 two axes is therefore perpendicular to 

 the radius. From this we can draw the 

 further conclusion, if we consider the pro- FIG 



perties of the ellipsoid that the common 

 radial axis may be an axis of the ellipsoid, and that, consequently, 



