44 Prof. J. Dewar .
The Boiling Point of Liquid Hydrogen , Thus , if we suppose that Mr. Forbes measured the total light within a circular area 5 ' in diameter , which seems a fair supposition , * the two measures of total brightness agree .
On the same supposition , the value of B ( B2- 1 ) in 1898 would be 1 3 full moon , and the total brightness of the corona would appear as M + 1*3 = 2-4 full moon .
Summary .
( \#171 ; .
) The brightness of the corona of 1898 at a point distant r from the sun s centre expressed in solar radii may be approximately represented by the formula brightness = Ar~6 + B , where A and B are constants .
( b. ) The first term may be considered as corona proper , while B may be taken as representing the constant illumination of the sky , or glare .
In 1898 the value of B was 2~6'4 = 0'012 moon , taking the brightness of the moon as 002 candle at 1 foot .
( c. ) The constant A varies with the radius along which measures are made .
In 1898 it varied from 20 0 moon to 21'9 moon , the mean being 2115 moon or 2*2 moon .
( d. ) The same formula will fairly represent the 1893 corona , the mean value of A being 2'23 = 1*2 , and the value of B 2~7'8 = 0-0046 .
( c. ) The total brightness of the corona depends on the area of sky included .
If a circular area 5 ' in diameter be included , the total brightness of the 1893 corona may be taken as PI full moon , agreeing with the visual measures made , and that of 1898 , on the same supposition , would be about 2-4 full moon .
" The Boiling Point of Liquid Hydrogen , determined by Hydrogen and Helium Gas Thermometers .
" By James Dewar , M.A. , LL. D. , F.B.S. , Professor of Chemistry at the Boyal Institution , and Jacksonian Professor , University of Cambridge .
Be-eeived January 8 , \#151 ; Bead February 7 , 1901 .
In a former paper f it was shown that a platinum-resistance thermometer gave for the boiling point of hydrogen - 238'-4 C. , or 34'-6 # The dimensions of the box are not given , either here or in the previous paper to which we are referred ; but on p. 369 of the \#163 ; Philosophical Transactions , A , 1889 , there is a diagram of the box , from which it would appear that the angular aperture was not greater than 12 ' , judging by outside measurements .
f " On the Roiling Point of Liquid Hydrogen under Reduced Pressure , " \#163 ; Roy .
Soc. Proc. , ' 1898 ( vol. 64 , p. 227 ) .
detevmined by Hydvoycn and Helium Gets 45 absolute .
As this value depended on an empirical law correlating temperature and resistance , which might break down at such an exceptional temperature , and was in any case deduced by a large extrapolation , it became necessary to have recourse to the gas thermometer .
In she present investigation the advantage claimed for the constant pressure gas thermometer over the constant volume thermometer is absent .
The effect of high temperature combined with large increase of pressure does not occur in these experiments , where only very low temperatures and a maximum range of pressure of less than one atmosphere were encountered .
At the same time , before dispensing with the effect of pressure upon the capacity of the reservoir of the thermometer , it was carefully estimated and found that it could not affect the volume of the reservoir by as much as 1/ 60,000th part .
This being determined , a particular advantage results from the use of the constant volume form , because in its case it is unnecessary to know the actual \ olumes of the reservoir , and of the " outside " space .
It is only necessary to know the ratio of these two volumes , and as this ratio appears only in the small terms of the calculation , it is not a serious factor in the estimation of such low temperatures .
Two constant volume thermometers ( called No. I and No. II ) were employed , in each of which the volume of the reservoir was about 40 c.c. , and .the ratio of the outside space to the volume of the reservoir was 1/ 50 and 1/ 115 respectively .
A figure of the apparatus is given herewith , where A is the thermometric bulb covered with a vacuum vessel to hold the liquid hydrogen , and be exhausted when necessary B is the manometric arrangement for adjusting the mercury at C to constant volume , and I ) is the barometer .
The readings were made on a fixed scale by means of a telescope with cross-wires and level attached .
A similar telescope was permanently fixed on the mark to which the volume had to be adjusted .
As the observations had to be made quickly , it was found convenient to use both telescopes on the same massive stand and to read the barometer placed alongside simultaneously .
The formula of reduction used was that given by Chappuis in the ' Travaux et Memoires du Bureau International des Poids et Mesures , ' tom .
vi .
p. 53 , namely , / v0(i + aT ) + ph \ 1 + 1 + ( H0 + h ) ... ( 1 ) , where Y0 is volume of reservoir at 0 ' C , T , temperature of reservoir , measured from 0 ' C , v , volume of " outside " space at the temperature of the room , t , temperature of the room , a , coefficient of expansion of the thermometric gas , 46 liof .
J. Dewar .
The Soilin Point oj Liquid EXHAUST / 3 , coefficient of alteration of volume of reservoir , clue to change of pressure , S , coefficient of expansion of substance of reservoir , H0 , initial pressure ( in these experiments always reduced to 0 C. ) , determined by Hydrogen and Helium Gas Thermometers .
47 H0 +h , pressure at-temperature T , after all corrections have been made .
On putting / 3 = 0 as already explained , equation ( 1 ) , by algebraic transformation and without any approximation , was altered into the form rp 273 + t + a ; 273 i \ Tl273 -M - xTi , ( Sa\gt ; ) ( 2 ) , where P - Po aP0 - SP ( 3 ) , in which Pq and P replace Ho and Ho + and v v0(l + aty The gases used as thermometric substances were hydrogen , oxygen , helium , and carbonic acid .
The values of a adopted in equation ( 3 ) were taken from Chappuis ' memoir , and were 0*00366254 for the first three , and 0*00371634 for carbonic acid .
The reciprocals of these coefficients are 273*035 and 269*083 .
The number " 273 " which appears in 6 is so nearly equal to the reciprocal of the former value for a that it was allowed to remain for the first three gases ; but in dealing with carbonic acid it was replaced by 269*083 .
In these experiments Tx is always negative , and numerically less than 273 , so that the value of 0 is always greater than unity ; nevertheless it differs from it but slightly , its value being unity when Tx = - 273 ' C. , and rising to 1*02 when Tx = 0 ' C. in the case of thermometer No. I , where x = 1/ 50 .
It may be noted that when is neglected Tx is the usual value given by Boyle 's law ; there is a convenience , therefore , in this form of Chappuis ' formula for approximation , because Tx can quickly be calculated , and the correcting factor 6 can be applied later if desired .
In the first experiment ( No. 1 of subjoined Table I ) thermometer No. I was filled with electrolytic hydrogen .
The initial pressure ( the pressure at 0 ' C. ) was almost three-eighths of an atmosphere , and was taken low in order to obviate any complication from condensation on the walls of the reservoir .
Two other possible causes might abnormally reduce the pressure at very low temperatures ; these were polymerisation and the presence as impurity of small quantities of gases liquefying above the boiling point of hydrogen .
The measurement of the density of the gas at its boiling point showed that there was no polymerisation , and further proof of this was evident in the constancy of the value of the boiling point when different initial pressures were taken .
To guard against the presence of gases with a higher boiling point than hydrogen , the electrolytic hydrogen was allowed to pass continuously for eighteen hours through the thermometric bulb before it was sealed off ' .
It was further calculated that an impurity of oxygen necessary to reduce the boiling point of hydrogen by a degree would amount to 3 per cent. , a 48 Prof. J. Dewar .
The Boiling Point of Liquid Hydrogen , quantity too large to escape detection .
This experiment gave the boiling point of oxygen as - 182'*2 , and that of hydrogen as - 253'-0 .
In the second experiment ( No. 2 ) a new thermometer , No. II , Avas constructed with a much smaller value of and as a further protection against the presence of impurities , palladium hydrogen was employed as the source of the gas .
A rod of palladium , Aveighing about 120 grammes , kindly placed at my disposal by Mr. George Matthey , F.E.S. , Avas charged with hydrogen in the manner described in my paper " On the Absorption of Hydrogen by Palladium at High Temperatures and Pressures , " * and subsequently used as the source of supply to fill the thermometer .
The initial pressure was slightly less than that in the first experiment ; the corresponding results Avere - 182'-67 and - 253'-37.t The neAV thermometer was filled afresh ( No. 4 ) Avith palladium hydrogen at an initial pressure rather less than one atmosphere , and gave for the boiling point of hydrogen the temperature - 252'*8 .
This result is a confirmation of the absence of polymerisation .
The next step Avas to compare these results Avith the results of similar experiments made upon another gas Avhose boiling point fell within the range of easily determined temperatures ; and as a further precaution the gas used in the thermometer was the vapour rising from the liquefied gas whose boiling point Avas to be determined .
The gas first selected Avas oxygen ( No. 5 ) , and as an additional condition to be noted , the initial pressure was made slightly more than an atmosphere , so that it would be in a Van der Waal 's " corresponding " state Avith the hydrogen in the first two experiments , namely , the initial pressure in each case was about 1/ 50 of the critical pressure .
The critical pressure of oxygen was taken about 51 atmospheres , and that of the hydrogen about 18 atmospheres .
There are good reasons for believing that the critical pressure of hydrogen is more likely to be about 11 or 12 atmospheres .
In the event of the lower value being eventually found the more correct , the effect as between the oxygen thermometer and the hydrogen thermometer will be to make the boiling point of hydrogen a little too high .
The result obtained from this experiment Avas to place the boiling point of oxygen at - 182'*29 , thus corroborating in a satisfactory manner the reliability of the method of determining the boiling point of hydrogen .
The question still remained , How far is a gas thermometer to be trusted at temperatures in the neighbourhood of the boiling point of the gas Avith which it is filled 1 To answer this question the oxygen thermometer was used to determine the boiling point of liquid air ( No. 7 ) in which a gold-resistance thermometer Avas simultaneously * 'Proc .
Cliem .
Soc. , ' 1897 .
f This thermometer gave 99''7 for the boiling point of water .
determined by Hydrogen and Helium Gas Thermometers .
49 immersed .
The gold thermometer had been previously tested and found to give correct indications of temperature down to temperatures not only well below the point in question , but lower than those obtainable by any other metal thermometer .
In the result the oxygen thermometer gave - 189'-62 , and the gold thermometer - 189^68 , as the temperature of that particular sample of air boiling at atmospheric ; pressure .
For another method of comparison this oxygen thermometer was partially discharged ( No. 8 ) until its initial pressure was nearly the same as that in the first hydrogen thermometers .
In this state it gave the boiling point of oxygen as - 182'*95 , establishing again the reliability of the method .
All the boiling points of the liquid gases were made on samples produced at different times .
As an extreme test of the method , I charged the thermometer Ao .
II with carbonic acid ( No. 11 ) at an initial pressure again a little less than one atmosphere , and used it to determine the boiling point of diy GCT the result was - 78'-22 , which is the correct value .
Hence it appears that either a simple or a compound gas at an initial pressure somewhat less than one atmosphere , may be relied on to determine temperatures doAvn to its own boiling point , in the constant volume gas thermometer .
Another thermometric substance at our disposal , as suitable foi determining the boiling point of hydrogen as hydrogen had been in determining that of oxygen and other gases , is helium .
The early experiments of Olszewski and my own later ones showed that pure helium is less condensible than hydrogen , and that the production of liquid or solid products by cooling Bath helium to the temperatures of boiling and solid hydrogen was only partial , and resulted from the presence of other gases undefined at the time the experiments were made .
The mode of separating the helium from the gases given off by the King 's Well at Bath is fully described in my paper on " The Liquefaction of Air and the Detection of Impurities."* If the neon , present as impurity in the Bath helium which was used , should reach its saturation pressure about the boiling point of hydrogen , the values given by this thermometer for the boiling point of hydrogen would be too low .
In order to avoid this , the crude helium extracted from the Bath gas was passed through aU-tube cooled by liquid hydrogen to condense out the known impurities\#151 ; oxygen , nitrogen , and argon .
In my paper " On the Application of Liquid Hydrogen to the production of High Vacua , " t it was shown that at the temperature of boiling hydrogen , oxygen , nitrogen and argon have no measurable tension of vapour , and that the only known gases uncondensed in air after such cooling were hydrogen , helium , and neon .
This same neon material # \#163 ; Chem. Soc. Pr^c .
, ' 1897 .
*1 " 4 Roy .
Soc. Proc.,5 1898 ( vol. 64 , p. 231 ) .
VOL. LXVIII .
E 50 Prof. J. Dewar .
The Boiling Point of Liquid , occurs in the gas derived from the Bath wells .
A sample of helium prepared as above described , which had been passed over red-hot oxide of copper to remove any hydrogen , was found by Lord Rayleigh to have a ref inactivity of 0*132 .
The refractivity of Ramsay 's pure helium being 0*1238 , and that of neon 0*2345 , it results that my helium contained some 7*4 per cent , of neon , according to the refractivity measurements .
This would make the partial tension of the neon in the helium thermometer cooled , in the liquid hydrogen to be about 4 mm. , and this being taken as the saturation pressure the boiling point of neon is about 34 absolute .
The initial piessure ( No. 9 ) was taken rather less than an atmosphere , and the temperature of the boiling point of hydrogen was given by this thermometer as - 252'*68 .
A further observation ( No. 10 ) was taken on another occasion with the .same thermometer , and the value found was \#151 ; 2o2 84 .
The fact that the boiling point of hydrogen , as determined by the helium thermometer , is in substantial agreement with the results obtained by the use of hydrogen itself is a conclusive proof that no partial condensation of the neon had occurred .
Of the remaining experiments in Table I , ( No. 3 ) was made in order to show the effect of a very small initial pressure , one-sixth of an atmosphere .
The results were unsatisfactory , owing to the sticking of the long column of mercury giving uncertain pressure readings .
In this case an error in the reading of a low pressure has six times as great an effect as if the initial pressure had been about an atmosphere .
If the temperature deduced for the boiling point of oxygen is corrected , , and the same factor of correction applied to the observed liquid hydrogen boiling point , then it becomes - 251'*4 .
It is of particular moment to have some estimate of how far errors in the observed quantities employed in Chappuis ' formula affect the nfinal value of T. In the case of an error in t , on differentiating equation ( 2 ) we get dT Ti \#151 ; a ; ( 273 + Ti ) ^ ( 273 + t - x\ff ( 4)- If x = 1/ 50 t = 13 ' Ti = - 180 ' : then dT = 000339\#171 ; 7 , or it would need an alteration of 2J ' in t to alter T by l/ 100th of a degree at the boiling point of oxygen .
In the same circumstances when It -riT = 0-00136 it , so that an alteration of between 7 and b mt e value of ( would only affect the boiling point of hydrogen by 1 ; 100th of From equation ( 4 ) the error in T varies with * very nearly .
Thus for the second thermometer where * = 1/ 115 , a variation of to the extent of 6 " , would only affect the boiling point of oxygen by 1/ 1Mth of a degree ; and it would require an alteration of U to affect the boiling point of hydrogen to the same extent .
determined by Hydrogen and Helium Gas Thermometers .
t-H rH 1 hH hH |m .
rM|H O |H * Carbonic Acid .
- tSjn IsSIliii 1 I 'm 1 t ' SO-* | o 1 u-5 M \#163 ; O H \#163 ; if W w X 1-lg1 1 12 1 11 !
1 1 1 I o *\#151 ; 1 h^ fa .
HH O |H fc it i l\#174 ; ca Dh '\#151 ; / 1 1*51* mi 1 1 1 '1 1 | !
1 l " 1 i X M O ^ \#163 ; P o bC s ?
o ID cn 9 9 d . ?
' i , , i i^i !
| Sis lg 1 s 1 1 1 's 1 1 6 HS fc 1 \#163 ; o b\#163 ; M* o ( M cp 1^isi isi i 1'i1 1 l\gt ; B 00 ( M r1 1 CO | M d HS \#163 ; p \#169 ; bO M o 9 9 d ?
, 9 , | , i | | | ; S'\#171 ; aS 1 S 1 1 1 1 S ' i ^ ^ x cq i ID M 6 J|\#163 ; ta=i p o b\#163 ; o iNlig " i *I ' 1 1 1 x cq 1 1 hH |m .
HH O H \#163 ; \#163 ; .2 So T$ O C3 r-j r-j 1'S IS 1 IIS 1 Nil 1 X\gt ; B t ' I CO p 1 , r*|H O H I a a 1 .2 8 , 1 P-j T3 1 PM Ph \#166 ; * ?
CO VC t\#187 ; oooi\gt ; \lt ; n \#171 ; \gt ; c eo cp |"lass is \#187 ; k|\lt ; | | 1 1 1 1 CM | hH hH |io 6 h \#163 ; \#163 ; \#163 ; O .rH bD ns O \lt ; 3 Jh p3h isiipia s Soil 1 1 1 1 i ^ W 6 HiS 1 ^ g .\#151 ; i \#169 ; 1 ?
n ^ fH O Q .h*h r -m \gt ; ?\gt ; M H O 00 co cc o *p ^9 iSlijsia i 's'S 1 1 1 i ^ !
i *1^ !
S " i % \#187 ; l O \#169 ; _s ^ co m Barometer Temperature of Rjom. . . .
Pressures at 0 ' C. B.P. of Carbonic Acid . .
\#187 ; Oxygen \#187 ; Air " Hydrogen " Hydrogen Solid ( 30 to 40 mm. ) Calculated Temperatures* B.P. of Carbonic Acid . .
" Oxygen " Air " Hydrogen ... ... " Hydrogen Solid ( 30 to 40 mm. ) / -\#151 ; s i. X ao Q Q .
~ " 8 52 Prof. J.Dewar .
The Boiling Point of Liquid In the case of an error in P , a similar process gives 0 ( oc - 8)P 273 + t ( aP0 - SP)2 273 xTi dT ( 5)- If x = 1/ 50 , t =13 ' , P , ) = 760 mm. , = - 180J ; dT = 0-3563 dP , so that an error of 1 mm. in P would only alter the boiling point of oxygen by a third of a degree .
In the same circumstances at - 250J , dT = 0-3516 dP , which is practically the same result at the boiling point of hydrogen as at that of oxygen .
For the second thermometer , these two equations become at - 180 ' , dT = 0-3575 dP , at - 250 ' , dT = 0-3548 dP .
In each of the last four results if P0 = ^x760 mm. the formulas become respectively dT = nx 0-3563 dT , and dT = n x 0-3516 dT = nx 0-3575 dT , and = n x 0-3548 dP ; in other words , any error in reading P is magnified in its effect on T directly in proportion as P0 is diminished .
This affords some explanation of the weakness of the results in Experiment ( No. 3 ) .
In like manner , from an error in P0 , we get ... ... ... ... ... ... ( 6 ) ' Here if x = 1/ 50 , t = 13 ' , P0 = 760 mm. , Tx = - 180J ; dT !
= \#151 ; 0*1188 dPo , or an error of 1 mm. in P0 would only alter the boiling point of oxygen by a ninth of a degree ; but with the same data at - 250 ' , dT = - 0-0264 dPo , so that the boiling point of hydrogen would only be altered by a tenth of a degree for a change of 4 mm. on an initial pressure of about one atmosphere .
In this case also if P0 = -x 760 mm. we get similar results to those 71 .
in the case of P , namely , For x = 1/ 50 , dT = -n x01188 clP0and dT - -\#171 ; x 0-0264 For 1/ 115 , dT - -n x0-1192 \lt ; \#171 ; P\#187 ; and dT x 00266 The general result of reliable experiments are an error in either P0 or P is , that the more those in which the initial pressure is as high determined by Hydrogen and Helium Gets as possible .
Hence Nos. 4 , 9 , 10 are in this respect the most reliable for hydrogen .
Also , it is of much more importance that P should be accurate than that P0 should be so ; in fact , for hydrogen an error in P has 14 times as much effect as the same error in P0 .
We can verify these results from Table I. In Experiment ( No. 2 ) , where P0 = 4 x 760 nearly , we have two readings\#151 ; one at the boiling point , the other in solid hydrogen , \#151 ; namely , 19*7 mm. and 14*4 mm , whose difference is 5-3 mm. This corresponds to = 3 x 0-3516( - 5-3 ) degrees , or 5'-59 .
The calculated temperatures for these pressures are -2530,37 and - 2580,66 , whose difference is 5'"29 , a satisfactory agreement .
If we compare Experiments Nos. 4 and 9 , in both of which the same value of a is used , we can pass from the former to the latter by the formula " 7T^ dT = - 0-0266 dP0 + 0-3548 P , in which dP0 = -11 mm. and dP = -0-5 mm. , whence dT = 0 -152 the observed result is - 252'-683 + 252'-806 or 0'-123 , which is also satisfactory and explains how so great a drop as 11 mm. m P0 has , nevertheless , so slight an effect on the result .
An alteration in the value of x has but little relative effect on the results .
As before we have dT T , ( 273 + 0(273 + Ti ) 1 ( 273 + t - sTi)2 CO- If x = 1/ 50 , t = 13 ' , then at W = - 180 ' , dT - - 57-085 at Tx = - 250 ' , dT = - 19-4205 and for the second thermometer ( x = 1/ 115 ) in like circumstances , and dT = - 57-895 dT = -19-802 For instance , if * were altered from 1/ 50 to 1/ 80 the result would be to raise the boiling point of oxygen by 0"-43 and that of hydrogen by 0''15 Finally , the alteration of a for any particular gas , being in any case small , affects the value of T practically only in its mam factor i- o hundredths of a degree therefore the change in 1 is inverse \gt ; p portional to the change in a , or , in other words , is directly proportions to the corresponding absolute zero .
, For instance , in Experiment ( No. 11 ) had we used the same value o a as for hydrogen the boiling point of dry CO* would have been - 79r-35 .
The Boiling Point of Liquid .
The following table shows what alterations would be required for each of .
the thermometers , in the values of t , P , P0 , and a ; to alter the boiling point of oxygen or that of hydrogen by 1/ 10 or 1/ 100 of a degree .
The table is calculated for t = 13 ' ; and in the cases of P and P\lt ; \gt ; the initial pressure is taken to be about 1 / nth of an atmosphere .
Table II .
Thermometer No. 1 .
Thermometer hfo .
2 .
Alteration of T. .
] r at B.P. of 0 . .
2 5 ' 6 ' 1 ' [ at B.P. of H . . .
7i ' 17 ' 100 J ' at B.P. of 0 . .
0-280 mm. n 0-280 mm. n 1 ' 1 [ at B.P. of H . .
0-285 mm. n 0 -282 mm. n Id i rat B.P. of 0 . .
0-842 mm. 0-889 rmn .
n n l0 H [ at B.P. of H . .
3-79 mm. n 3*76 \#151 ; \#151 ; mm. n to r j f at B.P. of 0 . .
0 * 88 per cent. 2 *00 per cent. i ' x s [ at B.P. of H . .
2-57 5 -81 " 100 Thus , for example , if the initial pressure in either thermometer were about half an atmosphere an error of 1/ 7 mm. in reading P would alter T by a tenth of a degree .
If we take the average values given by these experiments as being the most probable , then the boiling point of oxygen is - 182'-5 and that of hydrogen is - 252'-5 , or 20'*5 absolute .
The temperature found for the boiling point of oxygen agrees with the mean results of Wroblewski , Olszewski , and others .
If the boiling point of oxygen is raised to - 182 ' , which is the highest value it can have ; then an equal addition to the hydrogen value must follow , making it then - 252 ' or 21 ' absolute .
In a future communication the temperature of solid hydrogen will be discussed .
I am indebted to Mr. J. D. H. Dickson , M.A. , of St. Peter 's College , Cambridge , for help in the theoretical discussion of the results , and to Mr. Robert Lennox , F.C.S. , for able assistance in the conduct of the experiments .

