48 B. S. Woodioard — Iced Bar Base Apparatus. 



From the differences (1-2) and (3-4) of the table it appears 

 that the probable error of one measure of a kilometer is 

 ±0-T6 mm from the first series and ±0-09 mm from the second 

 series. The second of these probable errors, it may be re- 

 marked, agrees well with the corresponding value deduced 

 from the measures on the 100 m comparator, to wit, using the 

 average value for the probable error of one measure of the 

 comparator interval : 



,(30^ + 617) ,_ 

 ± i ,— VlO = ± O10 mm 



It will be of interest in this connection to compare these 

 probable errors of measurement with those obtained with 

 other micrometer microscope apparatus. The apparatus most 

 nearly comparable in this respect with the iced bar appa- 

 ratus are the Repsold bimetallic (zinc and steel) of the U. S. 

 Lake Survey and the Brunner bimetallic (copper and platinum) 

 used recently by French geodesists. The best work with the 

 Repsold apparatus, on the Olney base 1879, shows a probable 

 error of ±0°40 mm for one measure of a kilometer.* The 

 work of the French on the Paris and Perpignan bases, 1890, 

 1891, shows ±0'67 mm for one measure of a kilometer, f These 

 larger values are due probably to imperfect temperature indi- 

 cations of the bimetallic apparatus rather than to any material 

 differences of manipulation. 



An idea of the stability of the end and section marks of the 

 kilometer may be gained from the above table by computing 

 the differences between the mean lengths of the several sec- 

 tions resulting from the first pair and second pair of measures. 

 Thus we have : 



Mean of first pair minus mean of second pair of measures. 



Section m to 250 m — 0-58 mm 



" 250 " 500 +0-56 



'< 500 " 750 —0-50 



750 " 1000 +1-32 



These figures indicate considerable movements of the mark- 

 ing stones; and it seems not improbable that movements of 

 such magnitude did actually occur, since the stones rose to the 

 surface of the ground and the moisture in the ground varied 

 from the extreme of saturation to the extreme of dryness 

 during the interval which elapsed between the two sets of 

 measures. 



* Professional Papers Corps Engineers U. S. A., No. 24, p. 303. 

 f Comptes-Rendus des seances de L' Association Geodesique International du 8 

 au 17 Octobre, 1891, p. 182. 



