44 C. H. McLEOD ON NOTES ON 



The fluctuations in azimuth arising from changing temperature may be, to a great 

 extent, avoided by opening the shutters, doors and windows of the observing house several 

 hours before beginning the observations. 



"Where there is a sharp barometric gradient there wiU be a small error in azimuth intro- 

 duced through lateral refraction, but the error so introduced is very minute, and cannot, in 

 any case, be successfully avoided. 



Level errors. — There are a variety of circumstances connected with the use of the strid- 

 ing or hanging level in which small errors may occur. The level itself is frequently a very 

 imperfect instrument. Its chief defects arise from : 1st, lack of uniformity of scale value ; 

 2nd, imperfect mounting, giving rise, under changing conditions of temperature, to a distor- 

 tion of the level tube and consequent change of scale value ; 3rd, change of form and scale 

 value with age ; and, 4th, deterioration with age through the deposition of small quantities 

 of soda or potash on the interior surface of the glass, rendering the level quite unreliable. 

 All of these defects ma^^ be and, of course, are avoided by careful and experienced observers. 

 It is exceedingly difficult, however, to obtain a level, the scale value of which is perfectly 

 uniform, and which will remain so under all circumstances. The cliaiige of scale with age is 

 most elusive, and frequently gives rise to small constant errors. A level used by the writer 

 changed its scale value from 2-"19 in 1883 to 2-"56 in 1890. The change, as observed from 

 time to time, was a gradual one and appeared to have reached a maximum at the latter date, 

 seven years after its manufacture. In order to avoid errors arising from imperfection of 

 form in levels, Mr., now Professor, II. H. Turner, in the construction of the transit instru- 

 ments used in the Montreal longitude determination, attached the levels to their frames by 

 a hinge at one end and a micrometer screw at the other. When in use tlie liub])le is brought 

 to a central position and the micrometer read. The level is then reversed and the micro- 

 meter used as before. The diiference of the micrometer readings is then a measure of the 

 inclination. The method has proved a very satisfactory one, but requires some delicacy of 

 manipulation. The following is an example of the method, while at the same time the level 

 ends were read in the usual way : 



West End 18-2 East End 18-2 Micrometer 4-2 West 



" " 16-8 " " 20-1 " 4-2 East 



" " 18-2 " " 18-2 " 1-3 " 



Inclination from level readings-1-65 x •046=-^076 ; do. from micro. =-2-9 x 03 =-'^•087. 



If, for any given night in the work referred to, the results obtained from the micro- 

 meter method be compared with those from the ordinary method, it will generally be found 

 that the micrometer gives more constant values, or, in other words, there results from it a 

 smaller probable error than from the ordinary method. 



The most fertile source of level errors is, without doubt, in the deviation of the axis of 

 the level from iiarallelism with the vertical plane of the axis of the instrument. When a 

 cross level is attached to the frame, the error arising from this condition is eliminated by 

 causing the cross bubble to play at each setting. When there is no cross level, as is the case 

 with most instruments, the tendency is to bring the level standard next to the guard on the 

 transit instrument frame into the same position at each setting, and as this guard usually 

 secures the standard in a nearly vertical position the maximum error then occurs. 



