have been pursued, one by following the stream from the Gulf of Mexico, and making hourly observations of the temperature of the water, and the other by comparing the mean temperatures of the various sections with each other, and with the temperature of the Gulf of Mexico. In the first method, the vessel must be allowed to drift with the current of the stream, a difficult condition except in the best weather, even for a day, and to float along thus, for hundreds of miles would rarely be practicable. Any motion communicated by sails or by steam must carry the vessel beyond the water in which she commenced her voyage, and the lateral overflow carries the water constantly from the axis toward the edges of the Stream.

In the comparison of mean temperatures of the different sections, the fact has been established, that the temperature of the water of the Stream at any point may be higher than at a point nearer the source, and hence vessels in running along the Stream may, and generally do, pass through water not of a constantly diminishing temperature, but from cool to warm, and the reverse. This is to be explained mainly though not entirely, by the variability of temperature at the source.

By taking the mean temperature of any one section, and going back to the date of the departure of the waters from the Gulf of Mexico, as determined by the velocity of the stream, and comparing the temperatures observed with the temperature of the Gulf waters, it was supposed that a solution of the question might be obtained. The temperatures were taken from the most authentic meteorological records of the Gulf for a series of years, and those periods sought which corresponded to the dates desired. The uncertainty of the temperatures of the waters of the Gulf of Mexico, as obtained from air temperatures taken here and there along its shores rendered the results unsatisfactory. Enough seems to have been determined, however, to show that the surface temperature of the Gulf Stream along its course is variable; that a vessel sailing along the axis at a more rapid rate than the motion of the stream, will pass through water of higher and lower temperature, depending generally upon two conditions, viz: the distance from the Gulf of Mexico, and the temperature of the Gulf at the time the water entered the straits of Florida; and further, that the latter cause is the predominating one in the parts of the Gulf Stream adjacent to the Atlantic coast where the current is rapid.

The influence of the form of the bottom in forcing the cold counter current of the bottom upward, has been adverted to, and the fact appears to be well established in the cross sections where the ridges and valleys parallel to the direction of the stream separate it into bands of warmer and cooler water, and this conclusion, as has just been stated, is strengthened by the fact that the

bands and ridges simultaneously disappear south of Cape Florida. The phenomenon is moreover strikingly exhibited in the longitudinal section of the bottom, in connection with the lower temperatures.

The shallowness of the Stream in the strait of Florida, connected with the fact that the bottom falls off rapidly to the north and south afforded an excellent opportunity for testing the ques tion. If the cold water of the under polar current follows the bottom, it should appear in the shallow part of the strait, and here the warm water of the surface, and the cold water of the bottom, would approach each other. Diagram No. 1 shows the curves of 40°, 45°, and 50° (bottom temperatures) along the deepest part of the stream, commencing at Sandy Hook, and running as far as the Tortugas. All these curves rise with the bottom and pass over the ridge which divides the bed of the Atlantic from that of the Gulf of Mexico, and again fall with the slope of the bottom towards the Gulf. In the narrowest part of the strait where the depth is three hundred and fifty fathoms, the temperature from the surface to the bottom, ranges between.80°

and 40°.

On the effects of pressure on Saxton's deep sea thermometer.-In the exploration of the Gulf Stream, the temperatures below one hundred fathoms have mostly been determined by Saxton's metallic thermometer, and although the results have been consistent amongst themselves, and have agreed well with the indications of other thermometers, yet it was thought advisable to determine the effect of pressure by direct experiment.

Saxton's thermometer consists essentially of a compound ribbon of silver and platinum fused and pressed together by rollers. This ribbon is wound in a spiral form, one end of the spiral being firmly fastened to an interior solid axis and the other left free. Upon the free end is placed an index arm which moves over a círcular graduated scale carrying with it a friction hand or indicator which is left at the extreme point of the arc reached by the true index. The instrument is enclosed in a case to which the water is freely admitted. A variation of temperature is immediately noticed, as the effect is to give a rotary motion to the index.

The experiments to determine the effect of pressure were made at my request by Mr. J. M. Batchelder with means devised by Mr. Thomas Davison at the Novelty Iron Works. The following description of the apparatus employed, is given by the last named gentleman.

"The gauge consists of a brass cylinder H, about eight inches long into which a steel plunger is fitted, the upper part of the plunger at A being '70 of an inch in diameter, and the lower at B about 786, so that the difference in area of the ends is equal

to one tenth of a square inch. The cylinder is bored out a little larger than the plunger except for about a fourth of an inch near each end at C and D where both are accurately fitted. To the branch E a pipe connects, communicating with the hydraulic cylinder and leading the water into the centre of the gauge which it reaches after passing through the chamber F filled with sponge to prevent any impurities in the water from reaching the plunger. The upper end of the plunger connects by a wire W, to a spring as shown in the sketch at G, so constructed as to indicate pressure from 0 to 450 lbs., the spring being so strong that 450 lbs. produce a movement of the plunger equal to three-eighths of an inch. It is evident that as the difference in area of the ends of the plunger is one-tenth of an inch, one hundred pounds pressure from the water on this surface, as indicated by the balance, would equal a pressure of water of 1000 lbs. per inch, or a pressure ten times as great as that indicated by the balance throughout its scale. The only difficulty in the use of the gauge is that of fitting the plunger to the cylinder so that while it is perfectly free to move it is also perfectly water tight. This difficulty however has been overcome, and much advantage was also derived from Mr. Batchelder's suggestion for supplying the wear of the plunger and cylinder by depositing brass on the plunger through the galvanic process."

Connected with this gauge by a pipe is a strong wrought iron cylinder, sixteen inches long by four inches in diameter, in which the thermometer was placed, the opening being firmly closed by a screw plug. This second cylinder was immersed in a tub of water for the purpose of regulating the temperature. The thermometer once placed in the cylinder, is not again removed, the index being read by means of a mirror until the observations are completed. By the use of this apparatus, the effect of pressure up to 4000 lbs. per square inch was observed upon two thermometers, and the results are given below. The observations were made to indicate the effects of 500, 1000, 1500, 2000, 2500 lbs. pressure, etc. Seven series of experiments were made with thermometer No. 5, and five series with No. 10. The mean results show that a pressure of 1000 lbs. per square inch has no effect upon the thermometer; at 1500 lbs. the effect is less than one degree; and from 1500 to 4000 lbs. per square inch, the effect is to diminish the readings, the maximum effect being seven degrees.

The diagram exhibits the law of diminution by increase of pressure, and the depth corresponding to different pressures. The correction to be applied varies with the depth. For thermometer No. 5 it is only four tenths of a degree Fahrenheit at the depth of 600 fathoms. For thermometer No. 10, it is one degree at the same depth.

At 1500 fathoms the corrections are respectively five and a half and seven degrees.

Nearly all the temperatures observed in the Gulf Stream have been taken at depths less than six hundred fathoms.

Table showing differences of readings of Saxton's Thermomeler under pressure and free from pressure.

ART. XIX.-On the Chemical Composition of Pectolite; by J. D. WHITNEY.

A FEW years since I made some examination of specimens of a radiated fibrous mineral from Isle Royale, Lake Superior, which proved on analysis to be pectolite. A mineral, closely resembling pectolite, from Bergen Hill, New Jersey, which had been analyzed by L. C. Beck, and considered by him as identical with the stellite of Thomson, was examined at the same time and found to agree in composition with pectolite, as had been previously suggested by J. D. Dana. Both the stellite and Wollastonite of Thomson were referred by me, at that time, to pectolite, a reference the correctness of which has since been shown by Messrs. Heddle and Greg, in a paper on the composi tion of the English varieties of this mineral.t

Notwithstanding so many analyses of pectolite have been made. by different chemists, there has not been a sufficient accordance in the results obtained to justify a positive decision as to the real formula of the mineral, although that of Von Kobell has been generally adopted. It will be sufficient to refer to the various published analyses, to see that there is but an unsatisfactory degree of uniformity in their results, whether of specimens. from American or European localities. Thus, for instance, in * Journal of Boston Nat. History Soc., vi, 40.

+ Philos. Mag., [4], ix, 238; also in Erdmann and Marchand's Journal, lxvi, 144.

Von Kobell's analysis of the Monte Baldo pectolite, the silica is given at 51.3 per cent, while other analyses of Scotch and American varieties give as much, in some instances as 54 and 55 per cent of that substance. In the like manner, the amount of lime, as stated by different analysts, varies from 29.8 to 35.2 per cent, while there is even less agreement in the water, which is given at from 0:41 to 3.39 per cent.

The difficulty of procuring, in a perfectly pure state, a mineral which only occurs in a finely-fibrous condition is undoubtedly one of the principal causes of these discrepancies in the analyses; but it is also possible that the unusual care required for the correct determination of the silica in the very soluble class of minerals to which pectolite belongs may not, in all cases, have been appreciated. The great abundance and purity of the specimens of this mineral which have been obtained from the tunnel of the Erie railroad, recently excavated through Bergen Hill, seemed likely to obviate the first difficulty mentioned above. The results of three analyses indicated that this material was really of almost absolute purity, while no pains were spared to effect a complete and accurate separation of the various ingredients, and especially of the silica.

The pectolite dissolves more or less completely in chlorohydric acid, according to the strength and quantity of the latter. By using a considerable excess of rather dilute acid, all, or nearly all, the pulverized material may be dissolved into a clear liquid. As the attack is usually performed, a portion of the silica remains in solution and the remainder separates as a flocky precipitate.

The following experiments show the difficulty of estimating the silica correctly in this class of highly soluble silicates, and the necessity of unusual precautions in its determination.

On digesting the ignited mineral with acid until a perfect attack seemed to have taken place, the solution gelatinized on evaporation, and there was no perceptible gritty feeling when it was stirred with a glass rod or the spatula; on separating the silica, however, after evaporating to dryness, moistening with acid, and adding water, in the usual way, its amount was found to be equal to 62:10 per cent of the substance taken.

Another portion of the unignited mineral was attacked by acid, and the silica separated without evaporating to entire dryness; its amount equalled only 35.6 per cent. To procure the whole amount of this substance present in the mineral, and uncontaminated by any traces of the bases, it was found necessary to use the unignited substance for the attack with acid, carefully to evaporate to entire dryness over the water-bath, then to moisten the dried mass with strong acid and allow it to stand for some time before adding water and filtering. These precau