the ideal of ocular motion, though an ideal seldom or never realized in nature; or does the rotation of the eye subserve some useful purpose in vision? I believe there is no doubt that the latter view is the correct one, for there seems to be special muscles which are adapted for this rotation and the action of these muscles is consensual with the adjustments of the eye and the contraction of the pupil. This purpose I explain as follows. A general view of objects in an extended field is absolutely necessary to animal life in its highest phases, but an equal distinctness of all objects in this field would only distract the attention. Therefore the eye is so constructed and moved, as to restrict as much as possible both distinct vision and single vision. Thus as in monocular vision the more elaborate structure of the central spot of the retina restricts distinct vision to the visual line, and the focal adjustment still farther restricts it to a single point in that line; so also in binocular vision, axial adjustment restricts single vision to the Horopter, while rotation restricts the Horopter to a single line.

Conclusions.

The most important conclusions arrived at in this paper may be briefly summed up as follows.

1. The axial and focal adjustments of the eye are not so inseparably associated as is generally supposed; but on the contrary when distinctness of vision requires it they may be completely dissociated.*

2. In this dissociation, the contraction of the pupil associates itself with the focal in preference to the axial adjustment.

3. In optic convergence there is a rotation of both eyes on optic axes outward; and this rotation increases with the degree of convergence.

4. In inclining the visual plane downward the rotation of the eyes for the same degree of convergence decreases, until when the visual plane is inclined 45° downward_the_rotation becomes zero for all degrees of convergence. Below the inclination of 45° the rotation is inward. In turning the eyes upward, except in cases of strong convergence, the rotation also decreases slightly but does not reach zero ;t in strong convergence it increases as stated by Meissner.

5. Besides the rotation produced by optic convergence there is also a decided inclination of the vertical line of demarkation upon the horizontal line of demarkation, which increases with the degree of convergence. This change in the relation of these two lines is probably the result of distortion of the ocular globe.

* While these pages were passing through the press, I discovered that in this conclusion I had been anticipated by Donders and others. All previous experiments, however, were made by means of glasses. Mine were made with the naked eye.

+ See this statement modified in note on p. 165.

the

6. As a necessary consequence of the rotation of the eyes, for all degrees of convergence in the primary visual plane, Horopter is a line inclined to the visual plane, the lower end nearer the observer; but whether the inclination increases or decreases with distance I have not been able to determine with certainty. It probably increases with distance.

7. In inclining the visual plane below the primary position, the inclination of the horopteric line becomes less and less until when the visual plane is lowered 45° the horopteric line becomes perpendicular to that plane and at the same time expands into a surface. Below 45° the Horopter again becomes a line, but now inclined in the contrary direction, i. e., the upper end nearer the observer.

8. In inclining the visual plane upward or toward the brows, if the optic convergence be strong, the inclination of the horopteric line increases, but if the optic convergence be small it decreases but does not reach zero or become perpendicular.*

9. In looking downward 45° for all distances the Horopter is a surface passing through the point of sight and perpendicular to the median line of sight, but the form of the surface I have not attempted to determine. In looking straight forward at infinite distance the Horopter is also a surface passing through the point of sight, but the inclination of this surface I am unable to determine.

10. It is possible that in some eyes which would be considered normal there is, in convergence, a rotation of the eyes inward, probably from greater power in the superior oblique. In such cases the position of the Horopter would be different. Columbia, S. C., Nov. 16, 1868.

ART. XII.-Contributions to Chemistry from the Laboratory of the Lawrence Scientific School. No. 6.-On a new Salt containing Tin, Caesium and Chlorine; by S. P. SHARPLES, S.B., Assistant in the department of Chemistry.

SOME time during the month of November, 1868, Dr. Gibbs called my attention to the fact that when a solution of the chlorids of tin was added to one containing the chlorids of sodium, potassium, lithium, cæsium and rubidium, together with free chlorhydric acid, a heavy white crystalline precipitate was formed. Upon examination I found that this substance consisted almost entirely of a salt containing cæsium; this salt was so pure even from one precipitation that it gave the violet flame characteristic of the metal, and upon examination with the

* See this statement modified in note on p. 176.

spectroscope showed only faint traces of the other metals of the alkaline group.

The crude salt was dissolved in a large quantity of water to which some chlorhydric acid was added and the solution evaporated until crystals began to form on the surface of the liquid. It was then allowed to cool and the resulting crystals filtered off, washed with chlorhydric acid and dried. They were then found to be entirely free from even traces of the other alkalies. Two separate portions of the crystals were weighed out and treated with dilute nitric acid, which on boiling precipitated the tin completely as Sne2; this was filtered off and the chlorine determined as chlorid of silver. In the first analysis 4105 gram of the salt gave 1004 grams of Sne, and 5844 grams of AgCl. In the second, 4767 grams gave 1169 grams of SnO2 and 6727 grams of AgCl or in per cent

This it will be seen corresponds very nearly to the formula SnCs, Cl, so that the salt is analogous to the well known platinum salt PtCs, Cl.. Like this it crystallizes in the regular system.

In order to ascertain whether the salt might be made use of for detecting cæsium in minerals, three or four grams of Hebron lepidolite were fused with about their own weight of carbonate of soda and treated with HCl to remove the silica. To the strongly acid liquid a solution containing stannic chlorid was added; a slight turbidity was at once produced; upon standing a few hours a white precipitate settled to the bottom. The supernatant, liquid was then poured off, the precipitate washed with HCl and examined before the spectroscope; it was found to consist almost entirely of the cæsium salt.

The new salt will not serve for a quantitative separation of cæsium as it is slightly soluble even in strong chlorhydric acid. If the stannic chlorid is added to a neutral solution containing CsCl no precipitate is formed, but upon adding to the solution about its own volume of strong chlorhydric acid the salt is at once thrown down. Alcohol does not seem to have any influence upon the precipitation.

The double salts of the other alkaline metals with tin seem to be perfectly soluble in chlorhydric acid, rubidium perhaps the least so, but even this is not insoluble enough to interfere with the complete separation of the cæsium. As a method of preparing the salts of cæsium in a state of purity the precipitation of the metal in the form of chloro-stannate appears to present great advantages.

Cambridge, Dec. 31st, 1868.

ART. XIII.-Upon the Atomic Volumes of Liquids; by FRANK WIGGLESWORTH CLARKE, S.B.

If we divide the weight of a given bulk of any liquid at 0° by the weight of an equal bulk of its vapor at the same temperature, it is plain that the quotient will represent the number of volumes of vapor formed by one volume of the liquid, supposing both to exist at the above named temperature.

These quotients I term the vapor volumes of liquids. It is true that but few liquids can form vapor at 0°, and therefore these vapor volumes are fictitious quantities: yet, nevertheless, real or imaginary, exceedingly interesting results may be obtained by comparing them.

Of the relations existing between the vapor volumes of different liquids, I shall have but little to say in this paper, except in so far as they have been instrumental in determining atomic volumes. However, I will state as briefly as possible the leading results I have obtained by their comparison, but, as I am still at work upon the subject, I shall not enter into details.

In any homologous series of liquid compounds, the first member of the series possesses a higher vapor volume than the second, the second higher than the third, and so on indefinitely. I have yet found no exception to this rule, although I have calculated the vapor volumes of more than 400 different liquids. Furthermore, the difference between the vapor volumes of the first and second members of a series is greater than the difference between those of the second and third, and this again, greater than the difference between the third and fourth, and so on. The only exceptions I have found to this rule lie among compounds which stand so high in their respective series that the differences between their vapor volumes are very small, and consequently a very slight error in determining the specific gravities of the liquids is enough to account for the trifling variations from the rule. Even these exceptions are rare.

The first of these rules may be carried still farther. If we calculate the number of volumes of vapor actually formed at the boiling point by one volume of liquid at 0°, we shall find that the first member of a series forms more vapor than the second, the second more than the third, and so on as far as I have had data from which to calculate.

Hence it seems that the usual increase in the boiling points. as we ascend in a series, is not sufficient to counterbalance the decrease in the vapor volumes. Since, however, the differences between the vapor volumes are constantly diminishing as we

ascend in any series, it seems almost certain that there must be a point at which the increase in the boiling points will overcome the decrease in the vapor volumes, and therefore the amounts of vapor formed by the members of the series at their boiling points must begin to increase with every step upward. I have as yet found no such point, however, in any series upon which I have calculated.

In any series of substitution compounds, as a rule, the vapor volumes diminish as the atomic weights increase.

Chlorid of arsenic has a lower vapor volume than the fluorid. Bromids have lower vapor volumes than the corresponding chlorids, and iodids still lower than bromids. Sulphids, also, have lower vapor volumes than the corresponding oxyds.

Last of all I give the rule upon which most of my work upon atomic volumes is based. Whenever two compounds have equal vapor volumes, their atomic volumes also are equal, or nearly so, and, as a general rule, the greater the vapor volume the less the atomic volume, and vice versa. There is not an actual inverse ratio between the vapor volumes of liquids and their atomic volumes although a casual glance at the numbers would seem to suggest one. It must be remembered that the vapor volumes are calculated from the specific gravities of liquids at 0°, while the atomic volumes are referred to the boiling points, and therefore an exact inverse ratio would be very improbable.

Although as yet I have found no relations between the vapor volumes of liquids which would enable me to calculate their specific gravities at 0° from their composition, still the relations. which I have found seem to me to indicate decidedly the existence of definite relations between the atomic volumes of liquids, their boiling points, and their rates of expansion.

We now come to the subject named at the head of this paper, the atomic volumes of liquids. In 1855 Kopp published several articles upon this subject.* In them he described a method of calculating the atomic volumes of liquid compounds, showing that the atomic volume of any compound equalled the sum of the atomic volumes of the elements composing it, just as its atomic weight equalled the sum of the atomic weights of its constituent parts. For a large number of compounds he actually determined the atomic volumes, and thence deduced those of the elements contained in them; but, as he employed the old atomic weights, the numbers he gave for oxygen, sulphur, and carbon, have since been doubled. With this correction the list stands as follows. Hydrogen 5.5, chlorine * "Annalen der Chemie und Pharmacie," xcvi, 153 and 303, xcvii, 374, xcviii, 367.