Measuring and Recording Spectrum Bands. By T. Palmer. 279 
according to circumstances, which will be fully considered pre- 
sently. 
Fourthly. It is as well always to work with the micrometer, 
though it may not be your intention to use it ; it should be set by 
means of the adjusting screw to the position of some known band, 
and the division carefully recorded. 
The method I have adopted for measuring these bands is as 
follows : Firstly, as those which are symmetrical, or equally 
shaded to the right and left ; secondly, as those which are un- 
sy^nmetrical, and which are unequally shaded to the right and 
left. 
Of all symmetrical bands, it is best to measure directly for the 
centre, for, after a little practice, it is surprising with what exacti- 
tude the eye perceives the slightest inequality between the two ' 
halves of an object which in itself is quite symmetrical ; it is, 
therefore, I find, easier to place a division of the micrometer on 
the centre of any band at once, be it broad or narrow, than to 
estimate it by measuring from the two ends and so deducting. In 
the case of those very feeble bands of this class, it is advisable to 
move the micrometer alternately from left to right, and right 
to left, stopping when you have succeeded in getting a line 
exactly in the centre of the band. This movement of the scale 
will, I think, be found very serviceable in facilitating a correct 
result- 
When, as in the case of unsymmetrical bands, they are shaded 
throughout, that is to say from one end to the other, I measure the 
end which is the most marked ; and when the band appears nebulous, 
I take the centre. In case neither of the ends are distinct enough 
to allow of any definite accuracy in determination, or in the case 
of those very feeble bands which, in some instances, are scarcely 
visible, I take the apparent centre as near as possible, in which 
case the measuring is subjected to the following causes of error : 
Firstly. The eye of the observer is inclined to take for the 
cenft not the actual one, but a point which is more or less allied 
towards the maximum of the light. 
Secondly. When the luminous intensity, or the opening of the 
slit increases, the unsymmetrical bands are unequally enlarged, or 
drawn out towards the ends. 
Thirdly. Clear terminations, as in the case of all symmetrical 
bands, remain relatively fixed, whilst those which present a feeble, 
undecided appearance change from their original positions ; con- 
sequently the true centres become displaced. 
Fourthly. These centres, as well also as the whole spectrum, 
are materially changed by the thickness or strength of the sub- 
stance or solution under observation. This effect may be better 
x 2 
