Page images
PDF
EPUB

of air enters the result only as a divisor, with one-fifth its value. The measurements before and after explosion affect the numerator with one-third their value. All the errors of measurement are diminished threefold and fivefold in their effect on the result. The errors probable in the result with the Bunsen long eudiometer and with the Frankland and Ward apparatus are 0.027 and 0.019 per centum, for a probable error of a tenth of a millimetre in each scale reading. Therefore, while the Jolly apparatus can make more accurate single measurements than the other methods, it is not capable of obtaining so accurate final results.

If the apparatus be modified, as the air thermometer has been modified, by closing the end of the manometer tube, and dispensing with the barometer, the apparatus will give as good results as the Frankland and Ward apparatus. The probable error of a measurement becomes 0.014 per centum, and the probable error of the result 0.019 per centum, both on the hypothesis of a probable error of a tenth of a millimetre in each scale reading and of a hundredth at the Jolly-point.

The following table shows the probable comparative errors of the four apparatus mentioned.

PROBABLE ERRORS OF MEASUREMENTS AND OF FINAL RESULTS FOR PROBABLE ERROR OF ONE-TENTH MM. AND OF ONE-HUNDREDTH MM. FOR JOLLY-POINT.

[merged small][merged small][merged small][merged small][merged small][merged small][ocr errors][merged small][merged small][ocr errors][ocr errors][merged small][ocr errors][ocr errors][merged small][merged small][merged small][merged small][merged small][merged small][ocr errors][ocr errors][merged small][ocr errors]

SOME POINTS IN THE CONSTRUCTION OF AN APPARATUS FOR THE ACCURATE ANALYSIS OF GASES. By EDWARD W. MORLEY, of Hudson, Ohio.

[ABSTRACT.]

A. WHEN a certain method of determining the top of the pressure column is used, the conditions of most accurate measurement of varying volumes of gas are satisfied by making the column of mercury, which measures the pressure, equal in length to the part of the eudiometer which is filled with the gas. In the usual form of the apparatus, the pressure column has to be longer, which immensely increases the difficulty of securing uniformity of temperature.

B. To define the top of the pressure column, I use a Jolly-point enclosed in the barometric vacuum; this makes the probable error at the top of this column evanescent. The whole error of the tension is therefore produced by the same uncertainty which produces the error in the observed volume. If now the Jolly-point is put at about the level of the top of the eudiometer tube, the probable error of the tension is the same part of the tension as the probable error of the observed volume is of the observed volume. This is the condition of the maximum accuracy.

C. To preserve the barometric vacuum over the pressure column, I shut off this column from communication with the rest of the apparatus except at the instant of a measurement. The screw of the fine adjustment formerly described serves also this purpose. The vacuum is found to be even more permanent than that of a barometer.

D. An auxiliary pressure tube is employed for preliminary adjustment. The bore of this tube is but a millimetre, which secures two important advantages.

E. The eudiometer tube is graduated with fine lines not over the eight-thousandth of an inch wide, in half millimetres. The graduation is affected with no relative errors of the hundredth of a millimetre. A reading microscope is carried on a cylinder so solidly connected with the iron tripod of the apparatus that ten pounds produce a relative flexure of only a tenth of a millimetre. This microscope is brought nearly to the level of the mercury in the eudiometer when the column in the pressure tube reaches just to the Jolly-point. The microscope is made to give distinct vision

of the graduation, and by Grunow's cathetometer fine adjustment, the terminal lines of the eye-piece micrometer are made to coin. cide with two millimetres of the graduation. The focussing movement now moves the microscope so that distinct vision is had of the meniscus, and the level of the mercury is read to the hundredth of a millimetre on the eye-piece micrometer, whose divisions now represent the divisions of the tube carried forward into its interior. A four inch objective, with an amplification of about sixty diameters, is thus utilized.

F. The probable error of this reading of the level of the meniscus cannot be separated from other probable errors incident to the observation, but the sum total of all the errors whatever is equivalent to no more than a probable error of one-hundredth of a millimetre in this determination: and in the calibration of the apparatus, made before much practice in the use of this reading arrangement, the probable error of a reading was the hundredth of a millimetre.

G. The probable error of a single determination of oxygen in air is less than the four hundredth of one per centum.

H. The method of measurement is adapted to rapid computation. The whole reduction of three measurements in an analysis, so as to get the per centum of oxygen, takes less than two and a half minutes.

NUMERICAL RESULTS FOR THE MEAN RATIO OF OXYGEN TO THE SUM OF OXYGEN AND NITROGEN IN ATMOSPHERIC AIR. By EDWARD W. MORLEY, of Hudson, Ohio.

[ABSTRACT.]

SINCE the proportion of oxygen to nitrogen in the air varies. almost as incessantly as its temperature or pressure, the mean value can be accurately ascertained only by regular and continuous observation. The true value cannot be safely inferred from observations at irregular intervals.

A series of daily analyses in duplicate, continued for six months,

1

undertaken for a very different purpose, may therefore be of some interest as contributing to the knowledge of this constant.

The result is affected with errors depending on the following causes, the probable magnitude of which is well known.

1. The most serious uncertainty depends on the uncertainty in the calibration of the eudiometer. The probable error of a single determination at any given point on the scale is three milligrammes of water. At each point four determinations were made, so that it is as likely as not that the volume at any given point is known within a milligramme and a half of water. The probable error of the mean result, as far as this error depends on errors of calibration, is 0.0014 per centum.

2. The second source of uncertainty depends on the fact that the composition of the air is variable. Owing to this variation, the result of the number of samples analyzed is as likely as not to differ from the result from an unlimited number of samples by 0.0009 per centum.

3. The third source of uncertainty depends on the accidental errors of analysis. Owing to these errors, the mean result obtained from two analyses of each sample is as likely as not to differ from what would have been obtained by an unlimited number of duplicate analyses of each sample by 0.0002 per centum.

This result is also affected with one source of constant error which has not yet been taken into account. This error is due to the fact that the eudiometer tube, after the explosion, contains a larger volume of water than at the previous measurement. I do not see clearly how to determine the magnitude of this error.

It is at present hoped that other sources of error produce only results which may safely be neglected in comparison with these. The mean composition of the air, and the magnitude of the errors attending the determination, may then be stated as follows: Ratio of oxygen to sum of oxygen and nitrogen, 20.949 per ct. Probable error from imperfect calibration, Probable error from variation in ratio,

Probable error from accidental errors,

Total probable error,

0.0014"

66

0.0009 "

66

[blocks in formation]

.

SOME CONCLUSIONS AS TO THE CAUSE OF THE FREQUENT FLUCTUATIONS IN THE RATIO OF OXYGEN TO NITROGEN IN THE AIR AT

DIFFERENT TIMES. By E. W. MORLEY, of Hudson, Ohio.

[ABSTRACT.]

To study the subject above indicated, I have made duplicate analyses of air collected at Hudson, Ohio, on every day from January 1st to June 30th. The results have been graphically compared with the daily observations on temperature and pressure of the atmosphere made by the Signal Service at Cleveland, Ohio, from which Hudson is not far distant. Many of the variations in the amount of oxygen observed are closely connected with variations in the temperature and pressure. For the prediction or explanation of these, local observations of temperature and pressure might be sufficient.

I have also been favored, through the courtesy of General Myer, with the thrice-daily maps of the state of the weather for the period mentioned. From a comparison between these maps and the results of my analyses, which I have only this week begun, I derive some interesting conclusions, serving to confirm the notion that most of the variations in the amount of oxygen are caused by the vertical descent of air from above. I find this notion strikingly confirmed in some cases where it was some time since evident to me that Loomis' suggestion that the cold was caused by the descent could not be accepted as holding good. I find evidence that some depressions of temperature are caused by such descent, and at such times the amount of oxygen falls promptly at the beginning of the cold. But with the times of depression of temperature remarkable for their suddenness and severity, a vertical descent of cold air seems to be the effect and not the The descent follows the cold by a day or two or more, and at the time when the Signal Service maps lead me to suppose the descent has begun, and not till then, the fall in oxygen occurs.

cause.

The variations found this year have not been as great as those published by me last year, nor as great as those found by Jolly. I have therefore carefully re-examined those made before this year, and am confirmed in my confidence that they are affected with no unsuspected error.

[blocks in formation]
« PreviousContinue »