B'i-
.ILBERT F. 0 . G E R M = \ S X
The recent perfections in the experimental technique of the work on gases are so manifold and far reaching in their character that measurements of the highest degree of accuracv may now be made in one-half to one-fourth the time the methods of less than fiiteen !-ears ago required. This del-elopment is gii-ing rise to renewed actil-itv along these lines on the part of in\-estigators, new fields of work have been attacked, and progress made t h a t would have been impossible fiiteen years ago. The most conspicuous example oi this is the recent work on atomic and molecular weights b y purely physico-chemical methods, such, ior t-sample, as the revision of the atomic weight of nitrogen, undertaken and carried t o Yuccessful completion by Guy& and his collaborators. Since oxygen has b ! - common consent been chosen as the basis of all atomic weights, the accuracy oi ex-ery method wherein the density of this gas is in any way involved is limited by the accuracy n-ith which the density oi osygen is known. Until recently, howex-er, the densit!- of all gases was referred to t h a t of carbon dioxide free air,? weighed under the same conditions, and set equal t o unity. Hence the determination of any density presupposed a double operation, the consecutive weighing under the same conditions of air and tlie gas under consideration. Recent investigation has, hon-e\-er. shown t h a t the limit of accuracy attainable in the m-eight of a pure gas iar exceeds that attainable in the weight oi a i r ;
' Ph. A GLIYC"Recherche\ I3\pi.rimentales i u r le.; Proprieti., I'hy\icoChimiques d e cjuelques Gaz Reiirintetl from t h e "116Inoirci d e la Socii.ti. tie Physique e t d'Histoire S a t u r e l i e de Gen oi Lkcemher, 1 y o 8 . " S C J L I V ~ ~ ~ ~ ~ Recherches sur IC' I'oids .ltorniqui, d e l'.lzote." lecture deliveretl hefore thc Soci6t6 Chimique de I'aris in J u n e . 1gcl.j " T r a n u x 126cmt.; e x h i t w r la Revision tie.; Poitli .ltomicjuc;," .lrcli des Science.; 1'11~-. et S a t . tie GetiZve, 27, 5 5 7 r ~ c ) o q iZeit. : :itiorg. C h e m . , 64, I 1909 " S e e . for esarnple. A I.etitic. "12rchercht.s i u r le. G J L." 11. 2 ; 1 h 9 .\tin. chim. ~ j I i > - i, I 15, 2 j IS+ #. '
"
1;
and we have been forced t o conclude t h a t slight changes in the composition of the latter take place, depending on certain conditions not yet well established. It is interesting t o note, however, t h a t the earliest determinations of t h e density of gases were performed so as t o give the absolute weight of the unit volume; the results so obtained, however, were not strictly comparable, because there was no agreement as t o the temperature, pressure, or the units of measurement t h a t should be used to express the results, and the variety of units in use, not to mention the possibility of various interpretations, made a comparison a n exceedingly laborious undertaking. So, for example, T h . de Saussure? determined the weight of a liter of oxygen saturated with moisture a t 1 2 j and under a pressure of 7 j 8 mm. to be I 3563 g. Thomas Thompson’ calculates from this t h a t the density of oxygen referred t o air is 1.13j 2 1 ~Dumas and Bcussingault from apparently the same d a t a derive the value I 1oj6 From this circumstance arose t h e custom of referring all gas densities directlj- to air, for then there was no further chance of misunderstanding, and the results were intelligible t o e\-eryone. IYith the international adoption of the metric system for scientific purposes, this objection was no longer valid; b u t the custom continued in use, despite t h e objections of such great men as Regnault, and not until very recent times has the faultiness of the system been generally recognized. But e\-en the absolute weight of a gix-en \-olume of gas under gi\-en conditions varies from place t o place; and so, n i t h the refinement of balances, i t became necessary to define the place of weighing by indicating the latitude and the height abox-e sea level: in other words, to give the value of t h e gravity constant “ g ” a t the place where the measurements were
’
carried out : this again necessitates annoying calculations t o make the results of various experimenters comparable. The most modern usage is t h a t introduced bj- Morley, in which the density of a gas is the weight in grams of the normal liter, t h a t is, the weight of a liter of the gas under consideration a t the temperature of melting ice, under :i pressure of 7 6 0 mm of mercury, a t sea level in latitude 45' S :the expression "normal liter" will be used in the above defined sense in this paper. and designated by the abhreviatiori I-.,-.
Historical Survey Galileo in the first half of the seventeenth century had pro\-en t h a t air had weight; but the actual op:ration of n-eighing it presented unusual difficulties. Hon-el-er. the first measurements of the weight of air tvere made a little later 13>Otto \-on Guericke. after he had in 16jo in\-eiited thc air pump. Such early attempts must ha\-e been \ - c y - unsatisfactor!-, as e\-en the efforts of such great masters as I,a\-oicier. Berthollet. and Ilayj-, oi-er a century later, were still 1 - w ~ crude. La.-oisier make? ircyuent reference in his works to the densities of ox! il a n d air, without going into t h e cieti:ils of their measurenimt. 'Ihis is true of ixost oi the carlicxr experimenters, and. is 110'i to be tvondered a t , for it vas thouxlit t h a t all that \I as necewirj- for the detc-rmination ivas to n-eixli a flask. first empty, aiicl then filled with the gas in que;tion. noting the temperature and t h e height of thc lxxomccr- a L-ery simple operation indeed. E1-m the method described b!- Biot iti his "Trait4 de Ph!-sique,"' a method in u w for man?- ?-ears, n-as anythiiig but accurate. despite the author's recommendatioii. In its essential details, the method included the following points : -4 glass globe 0:' fil-e or six liters capacity, provided with a metallic stopcock capaiile of holding a Txcuum, is e\-acuated as ell as possihle, the r e ~ i d u a l pressure measured, arid t h c g:loi)e iveighed on a good balance. Then the stopcock i? don-!^- opened and the globe aliowed t o fill with air; the globe is again Ireighed, and the barometer,
' Hiot: Trxit& t l i
I'h?;-iciuc., I'ai-i., I, :,4;-3grometer are read Finally the globe is again evacuated and filled n i t h the gas whose density is t o be determined, and which has been stored over water. the globe is weighed. and barometer, thermometer and hygrometer readings again made T o arrive a t the density, a long series of corrections are necessary air displaced by the globe, cubical expansion of the globe, residual pressure, hygroscopic condition of the gas reduction to standard temperature and pressure Further Biot suggests t h a t the true weight of the globe is best expressed by the a\-erage of the weights empty, before and after the weighings full The quotient of the weight of air into the weight of gas under consideration gave the density relatil-e to the former The globe was calibrated b> means of water. and corrections made t o reduce to zero Then a simple calculation gave the absolute It is interestweight. which was reduced to sea level and 4j ing t o note that Biot recognized the danger of filling the globe with a gas saturated R ith water vapor for the gas. on expanding i n the evacuated globe, cools off to such an extent t h a t a large percent of the water vapor is precipitated on the walls of the globe, IT hich necessaril! results in a density t h a t is much too high Biot indicates t h a t this may be avoided by the desiccation of the gas. but adds t h a t this is less certain, and that by following certain precautions he believes no precipitation takes place ?'hat Biot erred is plainl! shown by the results obtained b! him and .lrago,' for the inexact 1-alues of the physical constants in use a hundred years ago nould not alone suffice to gil-e the uniformly high values they obtained 1-\ = I 299075 gm I2 \ = 1,43353 gm The method of Biot and Arago in i t \ essential details remained the itandard for a period of forty years, and the high esteem in which it was held may be seen from a remark made by Berzelius and Ilulong? in choosing a method for similar ~
(tl
work: . . . aprPs que 11. 51. Biot et Xrago eurent apport6 dans la mesure des pesanteurs sphcifiques des principaus gaz les soins les plus minutieus." These authors, howel-er. prefer weighing the gas dry, and recommend collecting it over water covered with a layer of oil to prex-ent impurities dissolx-ed in the water from contaminating the gas. and subsequently passing the gas o\-er desiccating agents. Sotwithstanding these precautions, they arrix-e a t a density for oxygen relati\-e to air much lower than the true \-slue. I . 1026, though it is in the sense that one would suppose; for Biot and -1rago had weighed oxygen saturated, or nearly so, with moisture. and air as it was furnished by nature. so t h a t the error on the latter must have been less than in the case of oxygen: this would lead to a higher relative result than when the two gases are weighed in the same state of humidit!-, or dry, as in the case of Berzelius and Ilulong. During the greater part of the first half of the nineteenth century the \-alms given b y Biot and Xrago remained classic: and this is not surprising, if we consider for a moment a list of \-alues of the density of oxygen referred to air found by the most careful experimenters, as gi\-en hy Ilurnas and Boussingault :' "
'
I
.os;
I
.OS8
I . 1026
I . loL: I . IO;-heating ordinary potassium permanganate in a vacuum They arrived a t the value
I'
\
= I 4202
gin
but belie\-ed the result t o be too high. Reference will be made in this paper to the probable cause \see page 47.1) The following >-ear. R . TI-. Gray3 during his work on the atomic weight of nitrogen found occasion t o determine the
weight of oxygen gas He used a globe of about 300 cc capacity, and a barometer measuring 2 0 mm internal diameter, connected up with a mercury pump as in Rayleigh’s work The gas was obtained by heating pure recrystallized potassium permanganate From Gray’s results, Guyel calculated the weight of a normal liter of oxygen t o be L\ o l ) s l , l = I 42596 gm In S e u c h Atel, Jaquerod and Tourpaian? worked along lines similar to those of Giesen , working with ordinary precision balances, they were able t o apply the principle of Archimedes successfully t o the accurate determination of gas densities By their method the! were led t o the value = I 4290
gm.
Guye. Kovacs and V-ourtzel) made thirty determinations of the den5ity of air in Geneva, and obtained the mean value L.\ = I 2930 gm The!- were led t o the conclusion already signalled by lIorley4 t h a t the density of air depends on the place, the altitude, and the barometric changes a rising barometer being accompanied by descending currents of air bringing supplies from the upper strata of the atmosphere less rich in oxygen and tending to decrease the density of the air, a falling barometer would have the opposite effect Table I contains all the more important 1 alues attributed t o the normal liter of okygen and air, of these, the x-alues found before Rayleigh’s time have little more than historical interest The 1-alue generally admitted today for the density of oxygen is The little differences found in its density can probably be traced t o slight impurities in the gas, the work n hich forms the basis of this paper was undertaken in the hope of adding 1
Jour chim p h \ \ 5 , 2 0 3 11907 Xrch 5ci p h \ s nat 1 4 3 1 , 2 0 1911 Comlites rentlu. 154, I j 8 1 ( r ~ r r Jour cliim p h \ \ \ m Jour x i 2 2 , 11j IiSSi
I O , 332 (1912)
something t o our knowledge of this point, and to determine the density of oxygen purified by the best methods a t our command. 'I'
I
iULC
Biot and .irago Berzeliui and Ihilong Dumas and B o u s i n g a u l t Regnaul t \-on Joll!
1
433 i3 ~
~
I
-Frc)3:,
I
qS92
Rayleigh
I
Norley J . Thornsen Lediic Jaquerod and I'intz'i
I
4'904 42tjoo 42906 42S76
GrajJaquerod anti 'l'ourpaian
I
1
I I I
4'92 42Sg6 .42()0
Guye, KO\a c i and D'ourtzel
-4s to the density of air, it can no longer he doubted t h a t slight variations in the composition of air do occur, and that its density must, therefore. not he constant: this question forms the topic of an excellent rex-ien- recently published by 0. F. Tower.' Two series of determinations of the density of air in Grne\-a were carried out I]!- the author after the completion of the researches recorded i n this paper, using the same precautions as those outlined. The results fully confirm the conclusions of von Jolly, Leduc, llorley. and Guye. The first series, three measurements, made during a period of rising barometric pressure, gave for dry, carbon dioxide free air,
I.\-
= I ,2027
gin.
corresponding to a deficit of os!-gen. if the a\-erage weight of the normal liter of air in Gene\-a be taken as 1.2930 gm. the a\-erage value found by Guye, Kol-acs and li'ourtzel. The second series, four measurements, made during a period of falling barometric pressure. gave : /-Y
\Vestern phys.,
11, 249
I
= 1.2032 gm.
Rvsvrvc. I.niver..ity
lgI,i'.
Hiillctin, 15, i , j S
I
rgrzi;
Jour. chitn.
corresponding to a deficit of nitrogen (see discussion on page 447) The work recorded in this paper x a s carried out in the laboratory of theoretical and technical chemistry of the Cniversity of Geneva, under the direction of Profesior Fh X Guve, whose interest in the progress of the work I take pleasure in acknowledging here, and t o whom, as well as t o Dr G Baume, I wish to express my thanks for friendly ad\ ice
Method The choice of a method t o be used is largely an arbitrary one, for each method in use has i t i own peculiar adxantages The globe method modified as it has been in Geneva is. lionever, relatively more rapid than the volumeter method since a single determination furnishes several values, and, as the value of an average depends in large measure on the total number of x-alues obtained, it \vas decided t o adopt thii method The most important modification which the method of Regnault had undergone in Geneva 15-as the siniultaneoui use of several globes of various sizes, requiring a greater number of weighings, but in all other points requiring no extra time or labor By thiq scheme sex-era1 important sources of error are controlled and practicallv eliminated, it ma! he ne11 t o refer to the most important of these Accidental errors of calibration are rendered imposiiblc if they be large, or unimportant if small T h a t large errors due to calibration are impossible was shown in a striking manner in the course of this work a large globe used in the meaiurements had been calibrated by 0 Scheuer. and subiequentlq used by J KoITacs in the course of his work on the density of air ,? a t its completion, the constants of the globe x-olume and contraction I v ere transcribed on a tagq and the whole itored away A4noxersight, howex-er, caused an error of 2 cc to l x made in the volume Consequently, thii globe g a \ e reiulti I
which differed by nearly a quarter of a percent from the results found with the other globes used in the work. A recalibration showed t h a t the volume was 872.33 cc, while the original notes of M r . Kovacs showed t h a t Mr. Scheuer had found 872.35 cc, a most remarkable agreement. Errors due t o condensation of the gas on the walls of the globes: since this error is a function of the surface exposed t o the gas. it depends on the volume of the globe, b u t is not directly proportional t o i t : for small differences in volume it can be shown t h a t
where s is the surface exposed t o the gas, and i' is the volume of the globe: and for the differences occurring in the globes used in this work, the numerical difference in surface exposed was approximately equal to one-third the numerical difference in s-olume By the use of globes differing materially in size, then, this source of error is reduced to a minimum, and an!- gross errors arising from it could be detected. As the cause of condensation is probably t o be traced t o the action of moisture in the pores of the glass, the error must diminish in proportion as this moisture is removed, arid the results obtained with different sized globes must theoretically approach the same value Errors arising from the adherence to the globe of extraneous matter, such a i moisture, minute particles of dust or slight deposits of chemicals readily formed in the atmosphere of a chemical laboratorj-, or t o the introduction into the interior of the globe of dust particles or other finely divided material. are reduced to a mintmum and made unimportant Small errors in the weights used are eliminated. An e\ample mill illustrate this. X set of platinum plated analytical weights made and adjusted by Scholl of Geneva, kindly loaned by C, Baume, was used in this work. The weights from one gram down were calibrated carefully, using the method described by Kohlrausch ; individual weights showed errors as high as one-tenth milligram. I n order t o test the necessity
of such accurate calibration, the final calculations of density were made in two ways. (a) Assuming the weights to be correct, and ( b ) applying the corrections found. T h e differences found by applying the two methods varied from zero t o I 3000, after taking the averages, home\-er, of each series of four globes, these differences fell t o from I 20000 to I 30000, or within the limit of error of the weighing5 second no less important modification was the method of purification of t h e gas-its liquefaction and fractional distillation a t reduced pressures. This is the only satisfactory way of preparing a perfectly pure sample of any gas, for by any other method traces of air are certain t o remain: traces which may be negligible, h u t as t o whose presence and quantit!there is no indication; hence the necessity of using a method of purification t h a t will eliminate all traces of air, as well as traces of any other impuritv. Gases more volatile than the gas t o be purified are gotten rid of in the first fractions which are discarded ; t h e less x-olatile products remain in the end portion, and are likewise rejected. Only those gases whose vapor pressure curves are approximately the same as t h a t of the gas t o be purified, or which form constant boiling mixtures with it, cannot be eliminated Psually i t is possible to purify from such gases by a chemical process, and this should in all cases precede t h e physical purification. innox-ation introduced into the method in this work is the simultaneous use of four barometers of different diameters, all set up in an air bath. The advantages of such a method can readily be seen, for by the use of a series of globe5 certain important errors have been minimized or eliminated, b u t the error due t o any uncertainty in the pressure has not been affected, so t h a t one of our greatest sources of error, and one t h a t has troubled all experimenters working on the densit!of gases, remained uncontrolled. As long as this great source of error remained, the accuracy of the other operation5 mas a useless precaution By increasing, then, the numhcr of barometers correspondingly t o the number of globes. we can proportionately decrease the error due t o a barometer
reading, and make this operation approach in its accuracy the operations already enumerated. Description of t h e Apparatus Experience has shown that an apparatus constructed entire]:; of glass can lie much more relied upon to hold a vacuum than one n-hose parts are joined together by ruhher or lead: and !-et the experimenter must he on his guard, for even apparently sound glass tubing may allow air to filter slowly through it. l'his is due to fine capil1arJ- openings running the length of the tube. rvhich. after fusion to the apparatus, may have one end opening on the interior, the other on the exterior. offering an ea.?)- pxssage to gases. Such imperfections are detected most easilj- hy nioistening the seals suspected of leaking with some highl!- colored liquid. taking care to have the apparatus evacuated. Xftcr some time the liquid will ha!-e filtered through tile defectii-e ieal, cc;lcring the glass on the interior . The apparatus used here was conrtructed entire1)- of glass : only in one place Irere rublicr connections used. namel) to attach the density g1ol-m. Hut c-x-en here the use of rubher is dangerous, nmre dang-erous, perhaps. than it TI-ould be in any other part of the apparatus: for no matter hon- careful one ma)- lie, there is alviays the chancc of introducing foreign niatcrial into the globes when attaching them to the rubber connections. A large majorit!- of the discrepancies in the results are probably to he csplained in this way. I t would seem highly desirable to replace the rubber connections used here by glass ; the flat glass joints used 11>- U70urtzel in the determination of the density of nitros!-l chloride' meet all the requirements. and the author is preparing to run a series of gaseous determinations with their aid. The use of the flat glass joints involves a little more risk of breaking the apparatus during a measurement, it is true, hut the increased accuracy will more than compensate for the added pains and trouble of manipulation.
The stopcocks, of which there were about thirty-fil-e on the apparatus in its perfected form, were all of the capillary \-ariet!-: they were the best obtainable : a small size, furnished b y 1ILiller of Liverpool. and a larger size. slightly inclined on the tube, made by Geissler of Bonn. In the preliminary work the?- n-ere lubricated with the rubber grease recommended Travers.' but it m-as found t h a t this grease slon-ly allsorbed oxygen particularly near the capillary opening%!There it came in contact with the pure gas; that this ahsorption gal-e rise to a change in the properties of the grease: and t h a t in a short time the air found an opening around the glass stopper to the interior of the globe, so t h a t a l-acuum could no longer be maintained. 2'he first and simplest remedy that suggested itself vas to modify the grease so t h a t it could 110 longer allsorb o:i>-gen. Since the base of the grease, rubher. is a n unseturated co:1ip(?u11d capable oi forming an addition product Kith an clement such as os!-pi; this was tricd. and prol.ed succ .A current of os!-gen was passed into the grease at a teniperaturc of about I j o ' C until no incre was ahorlied. The , p a w . rvhich n-as much more \iscous than it had been, retained its desirable properties, and prox-ed to b e a perfect lubricating i n 2 terial for stopcccks required to hold a \-acuum indefinitel!- in contact n-ith os! gen. I n fact, the stopcocks oi tivo of the globes used were greased with i t , and required no attention throughout the daration of the measurements, arid zt the end of fii-e months were still in perieci conditicn: the unmodified grease, 011 the other hand. had to lie changed e\-er!- two weeks. 'The stopcocks of the other two globes were greased c-ith the chlorinated grease used 11~7 \I-ourtzel in his work on nitrosyl chloride:? this grease has as its base stearine and paraffine, and is saturated with chlorine at 211 elel-ated temperature : it gave even better satisfaction thari the oxygenated grease, for it was less viscous, and the stopper turned more easily in its socket. Xs has been mentioned. a nen- feature of the method em1
Travers' Esperiinental Stud\- oE Jour. chim. p h y s , X I , 2 9 ( 1 9 1 3 ~ .
ployed was t h e use of a number of barometers equal to the number of globes employed. T h e barometers were of two types: the first, I and I1 in Fig. I , were of the model currently used in the Geneva laboratory; the second, I11 and 117 in the figure,'a special type recently described by the author I T h e chief point of difference between the two types was in the method of making and maintaining the s-acuum. ,Is shown in the figure, barometers I and I1 were connected up with a mercury pump, b y means of which the x-acuum was made and controlled, I11 and I V have their 1-acuum chambers (-i,.i) connected t o the capillary overflow tubes (b,bj and reserarranged exactly as the similar tubes in the voir tubes IAY,-Y)> mercury pump shown a t Z in Fig. I , making of them a kind of mercury pump. For a n accurate description of thc barometer, the reader must be referred to one of the reference. cited. The cleaning of the barometers, always a difficult b u t highly important operation, was accomplished as fo1~on.s They were filled with a strong solution of bichroniic acid and allowed t o stand over night, they were then filled with strong nitric acid containing a few drops of alcohol, and allowed to stand for several hours ; they were washed with distilled water free from grease, and finally rinsed with freshly distilled alcohol -After the barometers were placed in position, dry air was passed through them for a week to remove all adsorbed moisture. the>- were then allowed to stand ex-acuated for several days, and were filled without heating. For use in the barometer, mercury was specially purified Ordinary impure mercury was agitated, by bubbling air through it, for a number of hours with I O percent nitric acid; as an extra precaution i t was also passed through a mercury tower containing the same reagent. It was then distilled in a specialljconstructed still, composed entirely of glass Two ordinary distilling flasks, whose side tubes were fused to either end of a long condensing tube, serx-ed as distilling and receix-ing flask. The distilling flask had fused into its neck a tube 1
Jour kin C h e m
>(IC
3 6 , 2 4 j O 11914 Jour chim p h y s ,
12, 78 (1914
extending t o the bottom of the flask and terminating above in a Y, one branch of which was drawn out t o a fine capillary tube, t o admit a slow current of air during distillation; the other carried a stopcock and a thistle tube, through which
mercuq- was introduced into the flask The lower or receiving flask had a stopcock sealed in the hottom, to permit of drawing the distilled mercurJ- off without tearing don-n the apparatus; the neck was drawn out and sealed to drJ-ing tubes communicating with the water pump and with a mercur!manometer ; tn-o stopcocks, properly arranged, permitted
of interrupting the suction of the water pump and admitting air t o the apparatus The distillation was carried on a t about fifteen millimeters pressure, the small amount of air entering the distilling flask and bubbling through the mercury served the double purpose of preventing bumping and of oxidizing any easily oxidizable metali present, at the temperature of the distillation the mercury was not oxidized The method of operating the still will be evident from the description The barometers were mounted on wooden supports placed side by side, and these supports were used as the basis of a n air bath constructed about the barometers, as shown in detail in Fig. I , designed to prevent the rapid change< of temperature caused by an open door or window The plate glass front was composed of four sections g , and g2 Jverc stationary, the others, c j and g,, movable 50 as to admit of acc t o the stopcocks on the interior of the case In order to further guard the barometers from the temperature changes caused by the proximity of the ice baths about the globes during the operation of filling, two h e a l ? sheets of asbestos board were interposed between the barometer case and the globes ( )ver an opening j cm wide and the full length of each barometer left in the 11-ooden support behind each instrument was placed an accurately graduated plate glass scale These scales were qraduated on paraffine and etched with hydrofluoric acid b>- the ’ Soci6t6 Gene] oise pour la Construction deb Instruments de Physique The scales were dil-ided in niilli meters, and could be read to tenths of a millimeter b y the unaided e>e . b ~the - aid of a telescope the twentieth of a millimeter could be read 1% ithout difficult\- The internal diameter of barometer I ivas 25 m m , that of I1 wai 1,3 j mm 111 and I Y were 15 mni Each barometer uas accompanied by four carefull>- calibrated thermometers, one near the upper mercurjsurface. one near the center. and two a t the bottom. of the last t n o . one was on each side, to determine any difference in temperature in the two branches The thermometers were of the ordinary type$ with milk glass scale, graduated in degrees from 2 j ’to I 10’ The corrections for each thermom”
~
+
eter were determined a t the temperature of melting ice and a t the transition point of sodium sulphate decahydrate, thermometers showing abnormal readings were discarded. Barometer readings were made with the aid of a small cathetometer To facilitate the readings, an electric light immersed in a bath of water was arranged on a pulley hehind the barometer case, and could be brought into any position desired the blinding effect of the light was overcome ljj- interposing behind the glass scales a sheet of thin white tissue paper Both top and bottom of each meniscus were aln-a\-$read The corrected barometric reading. P,,, was obtained b! suhstituting in the formula I
where (I.,) is the ratio of the normal acceleration due to graI-ity to the acceleration a t the place of measurement; P ' is the obser\-ed barometric reading; i' is a constant equal to 6 . 6 ; I?, and 13) are the radii of curvature of the upper and lower menisci. respectively; T is the correction for the thermal e s pansion of the mercury and glass scale. The residual pressures in the globes were measured with the modified JZacLeod \-acuometer described hy G . Baume. The instrument was set u p in the barometer caw lxtn-een harometers I and 11. Its readings were verified b y comparison n-it h the barometers . The devices used t o obtain a \-acuum were the water suction pump. the mercury pump, the rotary oil pump, and the absorption of gases by amorphous carbon a t loti- temperntures. The mercury pumps, two in number (%. Fig. I were of the type described by Cardoso and the author.," One of them was reserved for use with harorneters I and 11. while t h e other served to evacuate the \-arious parts of the apparatus. : %
For t h e deri\-ntion and di>cussion of thi. formula \e? t h e a u t h o r . J o u r . Ani. C h e m Soc , 36, r+gO t 1 9 1 4 ) ;thi.: article alio contains a tal)le, giviiix the re5ult.; obtained \\.it11 the four liarometers in t w o wries of rwtliiig-i. 2 Jour. chirn pliys., 6, z G I 1908 1l)itl , IO, 306 , 1 9 1 3 ' . 8 .
45
Albert
F. 0.Geilnann
Evacuation with t h e mercury pump was always preceded b y evacuation with the water suction p u m p ; a vacuum of less than one-tenth millimeter was then attained in a short time. The suction pump was of a simple type, easily blown in the laboratory. To evacuate large volumes completely, the mercury pump proved t o be too tedious for this purpose the absorbent properties of amorphous carbon a t low temperatures were utilized. As is well known, certain forms of amorphous carbon have the property of absorbing large volumes of gas; this property becomes more accentuated as the temperature is lowered, while a t high temperatures i t ceases t o exist. Then i t is ex-ident t h a t t o prepare a sample of amorphous carbon for this purpose, i t must be carefully heated so as t o expel all dissolved gases. this also serves to drive out moisture and other volatile impurities, all of which would impair the absorbing qualities of the material. For this n-ork, cocoanut charcoal, as recommended by Dewar, was used, later investigation? indicates t h a t an'mal charcoal is superior It was enclosed in a tube, 7 (Fig 4 1 ,provided n i t h a stopcock, R-lqa t its open end, and sealed to the apparatus in close proximity to the volumes to be e\-acuated Plunged into a bath of solid carbon dioxide and alcohol, or into liquid air, i t was ready for use, and reduced the last fifteen millimeters of pressure due to oxvgen in a \-olume of two liters to one two-hundredth of a millimeter in fifteen minutes when liquid air was used as t h e cooling agent. The charcoal required frequent burning out to drive off the absorbed gases The rotary oil pump n-as connected as an auxiliarj- t o the mercury pump, but as the latter never failed t o give good results the oil pump was rarely used. It was of SiemensSchuckert manufacture, and easily ga\ e a \-acuum of I I O mm The globes in which the gas was weighed m-ere originall:~
I k n a r Cornptei rcndui 139, 2 0 1 1 r y q Chcrn S e i r i 90, jj 11y04 chiin p h ) i , ,XI 3, j (19041, Proc 1 2 0 ~ Soc 74, 1 2 2 I t g o j \ec also B a t r n a l t l Drude c l t 1 n , 23, S.+ 1907 I - Hempel anti 1 ater Zeit Electrochimic 18, 7 2 4 1 9 1 2
.Itin
ordinary round-bottomed flasks; they were drawn out at the neck and glass stopcocks sealed on. Two flasks were selected of the same glass and as nearly as possible of the same volume, and the smaller of them prepared as a counterpoise. The weight of the counterpoise was adjusted by introducing a little water so as t o be a trifle heal-ier than t h e densit!, globe. t h e volume was likewise adjusted to within a few cubic centimeters by blowing a bulb on the neck, and was then sealed Off
T h e method of calibrating a density globe is quite simple, and yet there are certain points t h a t must not he neglected if consistent1)- good results are to be obtained Fig. 2 ,
d Fig
2
in which the globes are sadly out of proportion, shoirs t h e form of apparatus used -4 is the globe whose 1-olume is to be ascertained, 13 is a stout round-bottomed flask of Jena glass The flask N was filled with distilled water, and the water allowed to boil for an hour t o expel all air. stopcock cl remaining open, and b closed, was then closed, a i d the was eT-acuwater allowed to cool off. Meanwhile the globe ated by means of the water pump, through stopcock i Xfter R had cooled somewhat, i mas closed, and a little water was allowed to flow into -A through stopcock b. Suction was again applied until a sufficient quantity of water had boiled away to sweep out all the air. Then b and ~1 were opened and A
was allowed to fill completcl!-. A4nydissolved gas remaining was boiled out by again lowering the pressure with the water pump. V7hen properly c!me. no air bubbles remained in the tube connecting the two globes. .A was surrounded with clean, finely cracked ice, and left until temFerature equilibrium was attained, keeping the water siphoned off. The connecting tubes n-ere removed irom globe A-l,its stopcock closed, and the capil1ar-J- above it dried m-ith a bit of filter paper. The again opened. and if, after half an hour, no mo\-emeniscus as observed, equili1:rium rras assumed to hai-e lxen attainec!, and the glolie was removed to the weighing room. after n-ipiiig it carefully. -4 short glass tube drawn out to a capillary was a t t w h e d to the stem of the glolie to provide for the expansion of the water; the globe itself was m q x n d e d from the balance arm 1):- means of an improvised basket made of nickel wire. l h c nest morning iveighings n-ere made to one ceiitigram. using counterpoises. Finall!the glolxs were ciiil)tiCd. e \ - ~ ~ . c ~ i a t earid d . weighed again. T h e difference n-as the ivcight of water contained a t the temperature of melting ice. '1'0 find the volun~c.at c' C the fol!on-ing corrections had to lic applied: I Correction for t h e liuoysncy of the air on t h e u-eights used. T h e aT-erage w i g h i of a cubic centimeter of air in Genc\-a is 0.001I 7 p i , tlieri the correction amounted to
''-
/. 1 1 - is the iveight of water. and S . j is 8.; x o . o c ~ ~ rwhere ; the correction is ncgatii-e. the density of br z ' Correction for the \-olunie of Trater a t zero: if the delisity oi water a t zero is taken a5 0.999S6S, the correction is 11- x 0 , ~ ~ y ~ ~ Yand h 0 ' ,ir, positi\-e. Correction for latitude anti altitude : in Gene\-a this is equal to + l 1 - X o . o o o o i h . ' The algebraic w m . ,: of these corrections !iecomes : 1:
+
= i / . ~ ~ o o 11. o o ~ i ) . o O o i . i ~ 11.
5~page 4;'
~-~
0.OOI 1T
8.j
io]- :I tliscuiiiuii oi thi, T,ictcr
11.
= O.DOOOZ/J
11.
or ten parts per million under the conditions outlined, therefore, the weight of the 11-ater at zero gives directly the \-olume a t the same temperature I n the following table are set down the 1-olumes found for the four globes used, which may he numbered. beginning with the smallest, I. 11, 111, I V , the values gix-en are the averages of a t least two calibrations for purposes of comparison the values found by ?re\-ious experimenters are also gi\ en ~
I
An important source of systematic error t h a t was neglected until Ravleigh called attention to it in 1S88' is the contraction of the globe when it is evacuated, causing it to dibplace l e s air than before, and hence apparently to neigh more To correct this error it is only necessary to knon- the amount of the contraction Its determination was a simple matter using the apparatus outlined in Figuie 3 The globe ita> enclosed in a stout metal can, supplied n i t h a lid that could lie clamped on tightly, the lid had an inch hole, pro1 ided with 1 ' two-holed rubber stopper, through one of the holes the \ten1 of the globe protruded I LI 1 , and through the other a cdlihrated The can n a s filled mith water, and the capillary tube W lid clamped on, so t h a t the water stood a t some point in the capillary the globe 11 as evacuated and the le\ el of the water in the capillarj- tube noted, then air was alloned to fill the globe rapidly. and the le\ el of the TI ater again noted The xolume of the capillary between the t n o points noted na5 the amount of contraction of the globe The a\-erage of i e i era1 determinations was taken as the final \ d u e The results are included in Table I11 -~~
_
_
Proc KO\ \m
43,
161-2
1
Chcm A C T \ \ 57, 74 i S h 8
I I1 I11 11-
I
2 j 2 001
410 2 7 455 7 2 8 7 2 33
15 6i I I8
'94
0 002
0
017
0 0001 0 OOjI
Gerrnann 11-ourtzel Germann Scheuer
Each globe and counterpoise was supplied with a nickel wire hook fastened about the stem, t o facilitate suspension in the weighing case, and in the balance case.
'- --'Up
I
I F1g. 3
The balance used in calibrating the globe5 was of a type used in technical work for weighing heavier objects, i t weighed accurately to less than one centigram. For the other weighing-- for the actual determination of the weight of the gase5an ordinary Sartorius chemical balance R as employed. To increase the accuracy, the pani were removed, and light
aluminium trays were suspended close under the knife edges by means of fine platinum wire: these trays were designed t o receive the weights, and left the lower part of the balance case free to receive the globes and their counterpoises. The center of gra\-it!- was lowered enough t o make the time of swing quite long, in order to facilitate the reading of'the sirings. The balance case itself was surrounded by a much larger glass case with doors opening from the front. Across the top of this outer case was stretched a nickel mire, from which the globes Jrith their counterpoises were hung preparatory to u-eighing. It was found t h a t this method of disposing of so man!- articles during the necessarily long process of weighing was much preferable t o the method usually used. namelj-, placing them cn a clean, smooth surface a t the hottoni of the case. The v,-ei::hts u-ere those used by G . Baume in his 11-ork they were ordinarj-, platinum plated annl!-tical ructecl by Scholl of Geneva ; they were carefully calibrated 1 2 ~ - the manufacturers and b y the author, 3 s has alread?- been mentioned. For the preparatiori arid purification of tht. os!-gen li\ed Ypecial loriiis of apparatus were emploTcd. The generator \vas cornposed of a set oi heal->- glass tubes, (,' Fig. 4 , u > n niunicating n-it11 one another. and filled n-ith the purest ;iotassium permanganate obtainable ' Kahlbaum I ; the gariate n-as co\-ered Jvitli a jieavy la!-er of glass wool tci 1-vtaiii any dust arising from the decomposition of the permaiiga:inte, There follon-cd a series of tubes contzining the rcagi;t.iit,sclcsigned t o effect the chemical purification of thc g::q: t!i comprised ;t tuhe containing solid potassium h!-drosidc i x tablet form to remol-e traces of czrbon dioxide ; t x o tulles containing phosphorous pentoside, to rerno\-e moisture : and a system of modified Liebig bulhs. I - , , containing merc.tir!-. to break up any ozone t h a t formed a t the temperature of decomposition of the permanganate ; a shunt tuhe xvith stopcock ___._
~
~~
im. ljhys., 6 ,
1
1goS8.
464
t
I
