OF THE UNITED STATES. 391 



fer to my operations witli the two feet theodolite as an ex- 

 ample, 



'I'hc methods thus decided upon will determine the num- 

 ber, as well as the lorm and coinhination, ol" the ohseivations 

 which arc required to give the greatest probability in the re- 

 sults, and these methods nmst be the constant rule for all the 

 observations. 



If oi)servations, which are limited by their nature to a cer- 

 tain time, are niade with repeating instruments, the number 

 of observations must be determined by a proper combination 

 of the theoretical formula or principle used in the case, the 

 nuv.iber of ol)scrvalions required tt) conqiensate the jjossiblc 

 errors of the instrument, a!id the accuracy aimeil at in the 

 observation. Tiiis is the case, for instance, in circummeri- 

 dian observations, tSjc. 



'I he quantity of work which is made in one summer's 

 campaign, the muliiplicity and variety of cases occurring, ob- 

 literatitig naturally the accurate remembrance of the pecu- 

 liar circumstances of an observation, which however deter- 

 mine tlie degree of relative confidence which it deserves, it 

 is necessary to be very accurate and minute in the notation 

 of every thing in the day books and journals kept of all ob- 

 servations and otiier operations. A certain systematic and 

 constant form must be observed in them ; they must be writ- 

 ten with such regularity and perspicuity, that any man, with 

 the proper theoretical knowledge, could execute the calcu- 

 lation"-, though he miglit have been unconnected with the 

 work itself. 



It is best to keep them in folio form, to bring as much as 

 possible under view at once, in order to facilitate the taking 

 out of the results for use in calculation, to admit room for all 

 accessory notices, and bring every thing readily under the 

 eye. 



If assistance enough could be had, it would be proper al- 

 ways to make two fair copies from the first journal, kept un- 

 der the direction of the observer. These first journals should 



s 



