has sustained some injury that has destroyed the collimation adjust- 

 ment. In the event of such extreme necessity, proceed as follows: 

 Bring the bubble to the center of the scale (mirror 



removed) and suppose the first reading to be . 5. 274 

 Loosen the collar clips and revolve the telescope on 

 its own axis a half turn, when the second reading 



may be 5.238 



( nrrecting for collimation error, while looking in 

 one direction through the telescope; the mean 



result will be 5.266 



Restore telescope to original position. Remove 

 ocular; revolve on vertical axis by use of release 

 key; adjust ocular to opposite end, level up a 

 second time and suppose the third observation 



tobe 5.242 



Turn telescope again upside-down and, pre- 

 supposing the collimation to be in error also 

 in this system, let it be assumed that the fourth 



reading is 5.274 



Correcting for collimation error, while looking in the 

 opposite direction through the telescope, the 

 mean result of the second set of observations will 



be 



The discrepancy shown in the two averages Indicates 



that the bubble is out of adjustment but a mean 



of the two averages, or 5.262 



will be a perfect result deduced by corrections possible in the instru- 

 ment itself by the infallible process of reversion an accomplish- 

 ment impossible with any other type of level heretofore constructed. 

 The instrument's first claim upon the attention of the engineer is 

 the facility with which the adjustment for parallelism may be tested 

 and the accuracy and speed with which it may be rectified. An open, 

 unbiased examination into those parts which might be suspected of 

 infidelity develops the conclusion ^that, one by one, they may be 

 reduced to negligible quantities. 



The influence of the newcryptic focus on minimizing the errors 

 of lens displacement on collimation adjustment is especially remark- 

 able. It is is discussed on p. 78. 



33 



