( 103 ) 

 2. Journey of 1904 June 2 — 23. Corrections of the chronometers. 



As this journey was made almost entirely by land the circum- 

 stances were much less favourable for the regularity of the chrono- 

 meters, in spite of all precautions taken. The intercomparisons, which 

 were made at least once every day, clearly show small irregularities 

 now and then and once on June 3, when the carriers were not 

 yet accustomed to their task, as I have said before, a serious 

 perturbation occurred. 



The instruments were almost always carried in the shadow of the 

 w r oods and only a few times they can have been exposed to the sun- 

 beams. We must assume, however, that they were subject to the 

 general fall of temperature which occurs in these parts in June, and 

 thence follows that we may not accept a constant rate for the chrono- 

 meter of Hewitt. 



From my regular thermometer readings in 1903 and 1905 I derive 

 for the mean fall of temperature during June 2°. 18, i. e. on an 

 average per day 0°.073 ; this would cause a variation of rate for 

 Hewitt of — S .018 per day. On this supposition and starting from 

 the time determinations at Chiloango of May 30 and June 24, the 

 daily rate would have been at the beginning (May 30 — 31) -\- s . 50 

 and at the end (June 23—24) -f S .08. 



As the temperature coefficient of Hohwü may be considered zero, 

 we may gather some evidence on this point from the relative rates 

 during the journey. Beginning after the perturbation on June 3, I 

 find the following differences between the two chronometers, each 

 being the mean result from at least three comparisons, and derive 

 from them the relative rates subjoined. 



Hohwü — Hewitt 



June 5 + 49. s 00 n iq7 



10 47. 17 _ "• ö ! 



15 44. 50 " " "• ^ 



20 39. 17 " " i" "« 



24 35. 50 " ' u " y ^ 



As the mean rate of Hohwü was about -|- 0. s 9, these values agree 

 fairly well with the assumed variable rate for Hewitt, which I there- 

 fore adopt as the most probable. For the middle of the period, June 

 12, the chronometer correction derived by means of the latter rates 

 differs from that which would follow from the constant rate -f- 0. s 29 

 by l. s 36. 



A great difficulty is caused by the perturbation on June 3, when 

 the difference between the chronometers seems to have varied abruptly 



