Rod-making. 241 



we have the conditions exactly reproduced a conical 

 plug fitting in a conical bore. Taking the bottle in the 

 left hand, and constantly turning it, tap the glass stopper 

 alternately on each side with any light piece of metal, and 

 in a few moments a cohesion which resisted all the tor- 

 sional strain you could apply, is so broken that the stopper 

 may be removed with the thumb and finger. 



Contrast with this the action of a perfect cylinder. 

 Insert it one-eighth of an inch, and it fits; insert it an- 

 other eighth, and it still fits, and a due proportion of co- 

 hesion is added to that already obtained; enter it fur- 

 ther, and still the same result each fractional advance 

 increasing the cohesion of the surfaces, until the limit of 

 insertion is reached. 



Now, whether a rod will throw apart or not depends 

 upon the relative proportions of the cohesion of the sur- 

 faces of the ferrules one with the other, and the centrif- 

 ugal motion imparted to the rod in the process of cast- 

 ing. So long as the former is in excess the rod can nev- 

 er throw apart. Start the " hour-glass " ferrule at all, 

 and the centrifugal motion preponderates. But the cylin- 

 drical ferrule may be withdrawn half an inch, and still 

 leave abundant cohesion to retain the balance in its fa- 

 vor. Ignorance or neglect of these simple and element- 

 ary principles have led to the construction of the " hour- 

 glass " ferrule, and to the claim that a simple ferrule so 

 made is liable to throw apart, I assent. But a simple 

 cylindrical ferrule is quite another matter, and when the 

 defects of the former are charged against it, guided both 

 by practical experience and theory, I insist that those 

 charges have no foundation in fact. Nor must it be sup- 

 posed that mathematical exactness of form or fit is essen- 

 tial to its practical success. I have known of a simple 

 16 



