somewhat the extreme ordinates; as we see that by simply

lengthening the

ordinates on the

northeast and

Southwest, the curve may be made symmetrical and its minor axis be brought into the merid

ian.

But it is evi

dent that, inanything which de

B

A

pends so directly upon the sun's heat, the average of all days will not give a true result. It is necessary to make a distinction between clear and cloudy days. From the records of the weather, the observations were therefore separated into two classes, viz: days more than one-half clear and days more than one-half cloudy; and there were found 35 of the former and 24 of the latter. The level readings of these two classes were then tabulated separately, and the mean readings taken as before, to represent the probable motion of any one clear day or cloudy day. The mean readings of the two tables, with the mean times of observation, were as follow:

These averages, with the signs changed, being made abscissas and ordinates, give the two curves marked B and C, the observations being indicated by numerals as before.

We see at a glance how necessary to a true result was the separation into two series. The two curves have indeed the same general form, but how different their sizes and positions. For clear days the longer axis is about 17", while for cloudy days it is not more than 4". The curve for clear days shows the same shortening of the ordinates that was noticed in A, though to hardly so great an amount. But the curve for cloudy days is very much distorted, and no longer resembles an ellipse. This excessive departure from the normal ellipse can not be attributed solely to the connection with the main building, for

any such retarding effect as has been supposed should be greater when the absolute amount of motion was greater, and therefore should be more evident on the curve for clear days.

An examination of the figure will, however, suggest another and probably the true explanation. All the points known in the curve under consideration, except the one given by the morning observation (marked 1), may be found upon a curve nearly similar to the one for clear days. And the distorted position of this point may be accounted for by the supposition (which in fact should not be unexpected), that on a cloudy day the sun would not affect the tower so early in the day as on a clear day. Consequently, when the effect begins to be perceptible, the sun has already passed the prime vertical so far as to shine very obliquely upon the northeast side of the tower, and therefore the tower inclines but very little to the southwest of its mean position, and the corresponding part of the ellipse is wanting. Again, the curve for cloudy days lies considerably farther south and east than the other, instead of being concentric with it.

As the south side of the tower is the one most affected by the sun's heat, and moreover has nearly twice the free altitude of the northeast side, it might naturally be supposed to expand most under the influence of the sun's rays; and this, combined with the connection of the northeast side with the building, and its inclination to the meridian, would throw the curve for clear days to the north and west of the other, as it is found to be. It is noticeable also that on clear days the tower is thus maintained out of its normal position by a considerable amount.

Still more, the mean inclination of the sun's rays to the vertical lines of the tower at noon, on the days classed as clear, was 21°, and supposing the intensity of the sun's heat to vary as the sine of the inclination, the minor and major axes of the ellipse should be to each other as sin 21° to sin 90°, or as 35 to 100. Now the minor axis is about 7", hence the major axis should be from this cause about 20". It is really about 17", a discordance which I think can not be considered great, when it is remembered that the tower, being more or less shaded by trees, etc., could not, even on a clear day, be affected by the sun, until the latter had risen some distance above the true horizon (for which the above proportion is calculated). and that the inclined position of the tower to the meridian would probably also affect this element.

The curve deduced from the observations is thus seen to exhibit a sufficient degree of correspondence, with what had been anticipated from theoretical considerations, to confirm the truth of the reasoning; the departures from the perfect ellipse being no greater than can be ascribed to the unsymmetrical position of the tower, its connection with the main building, and other disturbing causes. In all this discussion the varying

owers of absorption, radiation and conduction of the materials the tower have been left entirely unnoticed. They would ndoubtedly have their place in a complete investigation, but o data were at hand for estimating their influence.

If we may suppose the tower to be equally expanded in all arts of its height, the vertical side would be changed cA to a uniform curve, as AB in the figure; and since the eviation from a vertical is small, the tangent at A ould cut the vertical line BC nearly at the middle oint between B and C. Therefore CD is nearly one- D alf the height of the tower, and CA, the distance which he top is moved from its normal position, would be epresented by CD × sin CDA = height x sin of oberved change of level. This gives for the length of CA on n average clear day, 375 feet X sin 8"5 0185 inches, or .CA 037 inches, the major axis of the ellipse.

=

B

But for obvious reasons the upper part of the tower would robably be most affected by the heat; hence CD should proba ly be taken somewhat less than one-half the height, and the najor axis of the ellipse should be proportionately diminished. As an indication of the accuracy of the results obtained from uch a series of level readings as the above, it may be remarked, hat two days, which in the reduction of the tables were noticed is showing unusually large easterly and northerly inclinations, vere afterwards found, by reference to the weather record, to ave been marked respectively by a northeast storm and a cool orth wind. The effect of the cold rain or wind in causing ontraction of that side upon which it blew was thus plainly ecorded by the level, and a good illustration afforded of the ccuracy of which this method of investigation is susceptible. In regard to the practical bearing of this discussion upon the vailability of such towers for mounting astronomical instruents, it may be sufficient, without entering into details, to tate the conclusion arrived at in the present case. This was, hat the motion of the tower was so great and so uncertain as > make it unfit for the support of a meridian instrument, but ot great enough to seriously interfere with the differential heasurements for which an equatorial is principally used. The mean hourly change was not more than 1"-2, which, durng the few minutes of an observation, would be insignificant nless extreme accuracy was desired, and might be combined ith the unknown accidental errors in reducing the observations. In conclusion, I would express my thanks to Profs. Newton nd Lyman of Yale College, to whom I have been indebted for he use of instruments and for valuable suggestions in the conuct of the observations.

Brunswick, Me., July 11th, 1871.

ART. XXVI-On the destructive Distillation of Light Petroleum! Naphthas, at comparatively low temperatures; by S. DANA HAYES, State Assayer of Massachusetts.

UNDER the generic term naphtha, as applied to some of the distillates obtained in the arts from petroleum, is included a series of hydrocarbons having specific gravities above 0.742, or between 0.625 (rhigolene) and 0.742 (heavy naphtha), and boiling points varying with the densities from 65° F. to 300 F. These naphthas have distinguishing characteristics by which they are easily recognized and which place them in a class by themselves; and aside from their odors, densities, boiling points, volatility, and solvent powers, a noticeable peculiarity is the absence of oily bodies: they do not leave any permanent stain on common writing paper that has been dipped in them, as do all the heavier and oily distillates obtained from petroleum. The redistillation of these naphthas under different conditions produces other hydrocarbons, in which the proportions of hydrogen and carbon are not only changed, but some of these products are oils that will stain writing paper like fats, and it is possible to produce crystallizable paraffine from these volatile naphthas by properly conducted distillations.

In the summer of 1861, the writer had occasion to redistill several thousand gallons of light petroleum naphtha, which was entirely free from oily bodies, in cast-iron "stills," heated directly by coal fires and having powerful condensors attached, such as were then common in coal-oil refineries; and it was observed that besides the gases, light vapors, and a greatly diminished volume of naphtha, an unexpectedly large proportion of heavy paraffine oils was obtained; and after the distillations were finished, large masses of separated carbon were found in the stills, as in the ordinary destructive distillations of crude petroleum, or its heavy products.

In 1862,* Professor Bacon of Harvard Medical College observed, when examining a sample of "keroselene," a light naphtha, that had a specific gravity of 0.640 at 72° F., and when heated in a flask containing scraps of platinum foil, it began to boil at about 85° F. As the more volatile parts distilled off, the temperature continued to rise, and at 170° about three-quarters of the liquid had evaporated. It continued to boil freely, but the whole was not converted into vapor until the thermometer had risen considerably above 300°. It is remarkable that keroselene should be so readily and completely volatile at atmospheric temperatures. I found that keroselene and Squibb's ether, exposed in watch glasses, lost equal weights *This Journal, Sept., 1862.

in two and a half and three and a half minutes respectively; and the former evaporated completely in about two-thirds of the time required for the other.'

5 This peculiarity of petroleum naphtha has been so often observed in my laboratory, that I have learned to avoid the employment of heat when evaporating solutions, or extracts made in them, for the purpose of obtaining the substances dissolved; because, although these hydrocarbons are exceedingly diffusive, and evaporate entirely and very rapidly in the air at common temperatures, yet when heated above their boiling points (above 85° F., in the case of keroselene), they undergo destructive decomposition, or, if in a flask, destructive distillation, heavier oily bodies being produced which are difficult to remove from the residuum of such evaporations.

Within the past year an apparatus has been erected in Boston, by Mr. Z. A. Willard, for generating gases and hydrocarbon vapors for use in metallurgical operations, that has offered an opportunity to obtain considerable quantities of the oils made rom naphtha, distilled at temperatures below 300° F., and I have examined these products with much interest.

Willard's apparatus consists of three or more upright wroughtron cylinders, having a capacity of several hundred gallons each, standing near a common steam boiler, and which are connected together and with the boiler by pipes that enter at the bottom of each cylinder, ending there and starting out from the top of each again to connect with the bottom of the next; it is thus a system of large iron Woulfe's bottles, the first being connected with a steam boiler. These cylinders or gas generators, when in operation, are about half full of gasoline or petroleum naphtha of the lightest and cheapest kind, which leaves no permanent stain on note paper, while steam at common working temperature and pressure is passing in at the bottom of the first cylinder, bubbling up through and vaporizing the naphtha, then passing into the other cylinders with the same action. The cylinders are provided with glass tube guages, so that the changes occurring inside may be watched, and the whole apparatus and contents are maintained under a pressure of about fifty pounds to the inch when in operation.

In this apparatus the steam and naphtha vapors are held together in the upper part of the cylinders, above the liquid, inder pressure, and at a temperature of about 212° F., or much above the boiling point of the naphtha, but never so high as 300° F.; and the decompositions occur in the vapors rather than in the liquid, light uncondensable gases and vapors passing upward, and heavy oil falling down into the naphtha below. The apparatus was operated continuously by pumpng in naphtha at intervals as it was consumed, and after the