490 



SCIENCE. 



[Vol. II., No. 



it is believed, for general observation when less 

 accuracy is required, is shown in fig. 3. A 

 collar represented in section at a is inserted 

 into the tube of the spectroscope, and fastened 

 permanentlj' so that its front side shall be 

 just in the focus of the eye-piece. From the 

 lower front edge to the upper back edge, a silk 

 fibre passes, drawing back as it rises. The 

 fibre will evidently appear in the field of view, 

 as represented at &, as lines of diminishing 

 intensity. A set of horizontal, equidistant 

 spider -lines are attached to the front edge, 

 hence Justin focus. The 

 line whose intensity is to 

 be measured is made to 

 api^ear parallel and near 

 " riG. 3. * ^o O"*^ °^ '^^ interference 



lines ; and its intensitj' is 

 expressed by the number of the spider-Mne at 

 which the intensities correspond, counting 

 downwards. And here it maj' be mentioned 

 that such a scale of intensities (or, indeed, the 

 scale afforded \fy the micrometer screw read- 

 ings in the preceding apparatus) is not a scale 

 of equal parts, a change of a unit in case of a 

 line of high intensity being more than in case 

 of a low intensity. This is, however, believed 

 not to be a serious disadvantage in practice. 



The advantages of &j\j practicable method 

 of measurement over a mere estimation are 

 evident enough. When estimated by the e3'e, 

 it is believed to be impracticable to distinguish 

 more than five grades of strength, while by 

 this method quite fine shades of intensity can 

 be measured ; and what is, perhaps, of equal 

 importance, measurements made against dark 

 and light sk}' are apparently' identical, a 

 change in the brilliancy of the background 

 affecting the appearance equally of the ab- 

 sorption and interference lines. Evidentl}' an 

 unaided estimation would very likely be at 

 fault in such a case. 



As to the accuracy actually attained in 

 practice, it is found, in looking over the record 

 of about a mouth past, that the whole range of 

 the readings made at one observation, in ordi- 

 naril)' favorable weather, averages 0.3 of a 

 revolution of the micrometer screw ; and, as 

 from four to twelve or more readings are always 

 taken, according to the amount of variation 

 noted, the probable error of the mean may be 

 considered as about 0.03, as computation has 

 shown in a number of cases. Now, as the 

 whole range of the instrument used is from 1.0 

 to 5.7, it is evident that many grades of in- 

 tensity' are capable of appreciation. It is to be 

 remembered, that these readings are purposely 

 made in various quarters of the skj-, so that 



discrepancies in readings are partly due to 

 want of uniformity in the hygrometric state of 

 the atmosphere. It should be stated, also, 

 that such accuracy is not attainable below 2.0, 

 as the value of a unit is then considerably less 

 than above that value. 



The regular record made at the observatory' 

 is as follows : The ordinary meteorological 

 record is made three times dail^-. With the 

 spectroscope, at least three sets of readings 

 are taken, comprising measurements of the in- 

 tensity of the moisture line at the horizon, at 

 altitudes of 10°, 20°, 30°, and 90°. In all 

 cases, the readings are taken in all quarters of 

 the sk^' where there is sufficient light. A set of 

 readings is also taken by setting the microm- 

 eter at 2.0, giving a faint line just visible in 

 dark weather, and then measuring the altitude 

 at which the moisture line is of the same 

 strength. Such readings of altitude rarely 

 vary more than 2° to 4° in settled weather. 

 The strength of line and of the ' rain-band ' 

 is also estimated hj the &ye at each observa- 

 tion. At the same time the readings of the wet 

 and dry bulb hygrometer are taken, as well as 

 of a Regnault's condensing hygrometer. The 

 wet and dr}- bulb hj'grometer can be ventilated 

 b}' means of a bellows, as suggested bj' Mr. 

 H. A. Hazen, in a recent number of Science. 

 Notes are made of the direction and velocitj' 

 of the wind, of the clouds, and condition of 

 the air. 



One of the most interesting of the powers 

 of the spectroscope thus used is its abilitj' to 

 detect relatively' moist tracts in the atmos- 

 phere. While in settled weather entire uni- 

 formitj' at all points of the horizon is generallj' 

 noted, in unsettled weather considerable differ- 

 ences are often observed. An excellent ex- 

 ample of this power of the instrument occurred 

 on Maj' 26. During the morning when the 

 observations were made, the air was very 

 clear and dry, the moisture line therefore 

 weak. At 6 a.m., measurements made entirely 

 around the horizon showed that the line be- 

 came invisible very uniformly- at an altitude of 

 10°, except for about 45° of the north-eastern 

 horizon, where the altitude of disappearance 

 was 20°, while the iutensitj' of the line at the 

 horizon here was about double that elsewhere. 

 There was no wind blowing, and no clouds 

 of anj- kind were visible except a few wisps of 

 cirrus cloud high in the east. These facts were 

 all noted in the record at the time. At seven 

 o'clock, when the next readings were taken, 

 to my surprise tliis moist tract was found to be 

 nearly filled up with a bank of stratus cloud, 

 with no other clouds visible. At the same 



