zone-positive-electrized oxygen-in the atmosphere?" We see at once that the former question may be regarded as answered when the latter finds a satisfactory reply. Meissner considers himself warranted in assuming that what he has shown to be true in the combustion of phosphorus and hydrogen is typical of all processes of oxydation by atmospheric oxygen, and that, accordingly, this element suffers polarization in every instance where its affinities are exerted. "A bit of phosphorus with its immediate surroundings of air and water, in which it slowly burns, is a picture of the earth with its atmosphere and the oxydable substances, together with water, upon its surface: the white fumes which rise from the phosphorus (the solid matters held by them in suspension being disregarded) are not only chemically the same, as the fogs and clouds of the atmosphere, but the mode of origin of both is identical."-pp. 345-6.

Here we finish our imperfect analysis of this book: a book which we have studied with great satisfaction, both from the importance of the topics it discusses, and the philosophic spirit everywhere exhibited by the author. We are bound to say that the assumption of the formation of nitrite of ammonia from nitrogen and water is refuted by Meissner, and the true origin of the oxyds of nitrogen that occur in nature is satisfactorily explained. The ozone question by these researches acquires a broader basis and more consistent aspect than it has hitherto possessed, and the new fields of investigation that are displayed through these pages are full of invitation and of promise.

ART. III.—On the Transparency of the Earth's Atmosphere; by CLEVELAND ABBE.

BOUGUER in his Traité d'Optique and Laplace in his Mécanique Céleste have investigated the effect of the earth's atmosphere in absorbing the light of celestial luminaries. The latter has shown that, approximately, the logarithm of the intensity of their light varies inversely as the cosine of their zenith distance, and, more nearly, as the quotient of the refraction divided by the sine of the zenith distance.

If, then, assuming any unit of comparison, we observe the intensity i at the zenith distance 0, and the intensity I at the zenith distance zero, the intensity just before entrance into the atmosphere being put at i,, we have,

where Q is a constant to be determined, (e) is the density of

-

the atmosphere at the time and place of observation, and I is the height of a homogeneous atmosphere of the density (9).

Therefore we see that log E, varies with the barometer and thermometer. For the more accurate formula we have,

n2 .Q.1. 2K

where 80 is the observed atmospheric refraction, and -H a

constant equal to

This last formula depends on the assumption of a uniform temperature throughout the atmosphere, which assumption, says Laplace, can produce no great error. Therefore for the same place and standard height of the barometer and thermometer we should for all celestial luminaries have E, constant.

The numerical values of I and i, in terms of our assumed standard can be determined from two observations of i, and i, at the zenith distances, and 0.

To this end we have, adopting the approximate formula (1), cos 0, log i ̧ (cos 02 — 1) —cos 2 log i2 (cos 0, -1),

log I=

1

2

cos 021 cos 01

cos, logi,-log I

cos 02 log i2-log I

log i。=

=

1

(3)

(4)

Bouguer has given, on pp. 79-81 of his Traité, the result of two observations on the brightness of the full moon, which he made at Croisic in Bretagne, as follows:

the absorptive power of 7469 toises of air having the density of that at his place and time of observation. On the result of these two observations he bases the table on page 332, where we find that

which number is adopted by Laplace in the tenth book of his Mécanique Céleste.

As Bouguer's serial formula is not equivalent to our equation (1), I have repeated the computation by (3) and (4), and find

assuming of course the barometer to have been at 30 in. and thermometer 50° F. during the day.

On page 81, vol. xxxvi, of this Journal, will be found some observations which Mr. Alvan Clark has made, which will give

us a second determination of this constant. I was in hopes of securing a longer series of observations accompanied with a record of the barometer and thermometer, but these seem to be the only ones available for our present purpose.

Mr. Clark, by diminishing the apparent diameter of the sun until it is barely visible to the eye, thus determines its brilliancy in units of the brightness of a faint sixth magnitude star. However much his results may depend on the peculiarities of his method and the delicacy of his eye, yet we may fairly consider them as comparable among themselves.

We have then

1863, April 27, Boston M. T. 18h 40m 9, 73° 11'

783100 times that of the i=1308000 faintest visible star,

Whence we derive, assuming the barometer and thermometer to have been at standard heights,

I 1347632 E. = = 1680904

=+0.8017,

(C)

The difference between the values B and C may be considered as due to the diurnal and other changes in the heights of the barometer and thermometer at Croisic and Cambridgeport; to the differences in the transparency of the atmosphere and the annual and diurnal changes in the same; and to the fact that Bouguer observed the moon at night, and Clark the sun by day, therefore, the different laws of variation of temperature in the strata of the atmosphere and the brilliancy of the atmosphere as illumined by these luminaries will introduce discordances. This latter is the most important source of error; the light which we must measure in observations of this nature being the sum of those rays which penetrate the atmosphere plus the light of the atmosphere as illumined by those rays which it absorbs, In respect to this point it is very important that observations be made on the stars, comets and moon, as well as the sun. Mr. Clark's two observations at 18h 30m and 10m P. M. are interesting in this relation. They are as follows. Admitting a large circle of the illuminated air which surrounds the sun he finds, 18h 30m 0,=75° 0 i, 1055360 times that of faint02-28 20 41574400 est visible star. Treating these observations by equations (3) and (4) we find;

April 27th

28th

0 10

1

I 1606041

E。 = Fi1858843

+0.8640,

(D)

the rough agreement of which, with the values A, B, C, is due to the preponderating influence of those rays of the sunlight which penetrate the atmosphere.

It will certainly be interesting to make a more elaborate investigation of this subject and to examine also the transparency

of our atmosphere to the rays of heat and the chemical rays as well as to those of light. The importance of this matter in certain astronomical studies is not to be underrated; since we often notice the tendency to an erroneous estimate of the relative brilliancy of a comet's nucleus and its tail as seen through the evening twilight in the most interesting part of its orbit; also in investigations upon the form of our Milky Way based upon the number of stars visible to Herschel in his guages, or observed by Bessel and Argelander in their zones.

0

For the present we shall merely append a table of the computed values of E for the two values of E, given in (B) and (C), together with the corresponding values abstracted from the table given by Bouguer, p. 332. The 2d, 3d, and 4th columns give the E which results from assuming i=1. The 5th column contains the quotient for the value (C) of E。.

E

Eo

ART. IV. On the Distribution of the Dark Lines in the Spectra of the Elements; by Prof. GUSTAVUS HINRICHS, Iowa State University.

As soon as I heard of the great discovery of Kirchhoff and Bunsen, I felt sure that the dark lines of the elements would prove to be distributed according to simple laws, and that these laws might lead us to a knowledge of the relative dimensions of the atoms. But it was only quite recently that I was enabled to study the distribution of these lines in its detail, Prof. Silliman Jr. having kindly sent me Kirchhoff's two memoirs. Yet I limited my investigation to the group of the alkaline earths and iron; for whatsoever is true for some elements will also be found applicable to the remainder of them; and besides Kirchhoff's measurements are not at all to be considered as final, neither does he give all lines that may be observed: so that the material at hand can justify only a preliminary investigation, to be completed and modified by more complete and more accurate observations.

As it is well known that some distinguished American experi menters are engaged in a more accurate and complete survey of the spectra, I thought it proper to give the results of this preliminary research in order that the remarkable laws here found as the essence of Kirchhoff's determinations might as soon as possible be further developed.

It might seem that such an investigation could not lead to definite results, since Kirchhoff's scale is entirely arbitrary and even changing. But within a small range equal differences on this scale must correspond to equal differences in wave length, for whatever the curve may be (we have found it to coincide with a parabola), representing the wave-length as a function of Kirchhoff's millimetre-scale, within such narrow limits the curve will coincide with its tangent. We must therefore first see whether the different groups of lines exhibit any order.

Commencing with the Calcium-spectrum, we find a group of lines near Fraunhofer's G. Representing the intensity by I, the scale-division by K, the successive differences by D, we find from Kirchhoff''s table

It is plain that the values of D are as 1:2:4; considering these ratios as intervals, i, and calculating from them and the difference the values C, which would correspond to a perfectly regular distribution of the lines, we obtain the error of this theoretical determination E=C-K, as given in the last column. We see, the error is almost nothing-the last one might easily be conceived as the result of the increased range. Hence we find that this group is very regular, having a principal difference of 50, or a simple multiple (here successive duplication) thereof. The second group, stretching from 2653.2 mm. to 2605-8 mm., has a range of 474 mm., which appears to be rather large; yet we find (comparing lines of high and similar intensity, those of lower intensity not being uniformly given):