]\gt ; .
M. W. Travers and Mr. A. G. .
Gwyer .
[ Jan. 11 , Professor Judd has enabled me to try the action of radium emanations on specimens of this glass .
Fifteen milligrammes of pure radium bromids sealed in a quartz tube were put close to a sample of the deepest green glass .
In two days appreciable darkening took and in 10 days the darkening was very marked .
A similar experiment was tried on a sample of the purple-tinted glass ; here also the action was the same as in the former case , and in 10 days the darkening was about the same .
Allowing for the apparent alteration of colour caused by tint of the two pieces of glass , I should say the colour produced by radium was similar in each case , and was of the same purple colour as that caused by long exposure to sunlight .
1 ' On the Comparison of the Platinum Scale of Temperature with the Normal Scale at Temperatures between and C. , with Notes on Constant Temperatures below the -point of Ice By MORRIS W. TRAVERS , D.Sc .
, , and A. G. C. GWYER , B.Sc. Beceived January 11 , \mdash ; Read January 26 , 1905 .
Infroduction .
the last few years one of us has been engaged in researches involving the accurate measurement of low temperatures .
The measurements were made by means of gas thermometers , of the constant volume type , filled with hydrogen , or with helium , at an initial pressure of 1000 mm. As , however , this method proved cumbrous and inconvenient , it was decided to standardise a platinum resistance thermometer of the type against the gas thermometer , and to employ it in future investigations .
The comparison of the thermometers between and C. has been carried out with considerable accuracy , and as the range of temperature overlaps that investigated by Callendar and Griffiths , and by Harker and Chappuis , we have decided to publish resuIts , though from our own standpoint the work is only considered as preliminary to other investigations .
In expressing our results we have employed Callendar 's notation .
Thus , when is the resistance of a coil of platinum wire at ths melting point of ice , and is its at the temperature of saturated steam under absolute standard pressure , which in the laboratory of University College , Bristol , is equivalent to mm. of mercury at zero , the value of the coefficient of change of , is iven by 1905 .
] On the Scales of Temperature .
If the coil has a resistanoe at the platinum temperature .
Callendar and Griffiths have shown that between and the boiling point of sulphur on the constant pressure nitrogen scale , the difference , , between the platinum and the gas scales can be calculated by means of a parabolic }'mula , ; where is the temperatnre , on the gas scale , and a is a constant .
Callendar hns sllggested that this principle form the basis of a practical scale of tempernture , and hIessl'S .
Johnson and Matthey have taken much lble to pure the purpose .
For this material is or the valus of about a set of platinum thermometers , hich the value of was about with the constant ogen t have found the diffsrence } ) etween the two scales of temperature can be calculated by means of Callendar 's formula , for ) .
The comparison carried out over a rallge of between C. and C. , the errors at the higher temperatures did not exceed More recently , Harker extended this investigation , comparing the platinum thermometers constanf nitrogen thermometer standardised at , the tter being for the boiling point of slllphn on the air scale .
The observations , therefore , really refer to the constant pressure scade .
In this research he employed resistance thermollleter described the British Association Report lS99 ( p. 243 ) , which the of is .
Taking for the value of , he found that the temperatures calculated from observations of the two thermometers showed a maximum which only exceeded the probable } of measurement over the higher part of the range .
At the end of his pnper Harker gives a table of the corresponding values of and , calculated on the ) asis , between and 1 , .
As it is well-knowl that the formula cannot be applied to the calculation of the difference between the two scales low temperatures , this to be lmnecessary and tding .
Sevel'al independent invcstigators tinum thermometers at low turcs , but for the reitson that they did not * ' Phil. , 1891 , p. 119 .
'Phil .
Trans , 1900 , vol. 194 , p. 37 .
'Phil .
rans .
vol. 203 , p. 343 .
Dr. M. W. Travers and Mr. A. G. C. Gwyer .
[ Jan. 11 , employ standard platinum , or did not succeed in maintaining sufficiently steady temperatures , the resuIts are not strictly comparable with those already referred to .
The results of DelVar and Fleming , and of Olszewski have been summarised by Callendar ; * they show varying divergencos from the parabolic formula , depending on the kind of platinum empJoyed .
Holborn and Wien compared the platinum , air , and hydrogen scales at temperatures between and , and again at and at .
They employed a platinum coil enclosed within the gas thermometer bulb , so as to minimise the error due to varying temperature .
They found with -platinum for which was equal to the parabolic formula gave a very close approximation between and , but showed a deviation of at .
We find that the deviation becomes marked even at Apparatus \mdash ; The thermometer consisted of a coil of platinum wire about 130 cm .
long , having a resistance of 7 ohms at the ice point .
The ends of this wire were fused to leads about mm. in diameter , and these leads and the ends of the compensating leads were soldered respectively to the ends of four insulated copper wires enclosed in a lead tube .
The resistance was measured by means of resistances constructed by the Cambridge Scientific Instrument Co. , to the design of Professor Callendar .
The individual resistance coils were carefully calibrated by one of us before the experiments were commenoed .
As the coils were of platinum-silver alloy , for which the temperature coefficient , it was necessary to observe their temperature each time an observation was made , and to reduce the result to zero .
The resistance )lugs were cleaned before each experiment , and the zero point on the sliding resistance , which rarely varied by more than ohm bstween individual sets of experiments , or by more than ohm in the course of two years , was always observed .
A home-made galvanometer of the two-coil astatic type was employed , and with the highest resistanoe in the circuit it was possible to adjust the slider with an accuracy corresponding to ohm .
In connection with these instruments the usual arrangement of keys , etc. , was employed .
The undamental Interval of the \mdash ; In determining the ice and steam point the platinum coil and leads were enclosed in a glass tube containing some paraffin to the height of about 3 cm .
above the coil .
An identical resuIt for the ice point was obtained by protecting the thermometer by means of a glass tube with holes in the side of it and immersing it directly in melting ice and water .
The following are the resuIts of the experiments , which need not be given in * ' Phil. Mag February , 1899 .
'Wied .
Ann 1896 , 59 , p. 213 ; 1901 , 6 , 1905 .
] On the Platirv um Normal Scales of Temperature .
531 measuredIcepoint .
Steam p Date and place .
July , 1903 , London Three times Twice October , 1903 , London , , Once July , 1904 , Bristol , , Twice The probable value of the coefficient is , therefore , about It may be pointed out the maximum variation in the ice point would correspond to one unit in the next place of decimals .
The variations in the experimental values of are entirely attributable to errors in the determination of the steam point .
In any case an error of one unit in the last place of decimaIs is insignificant and amounts only to an error at Comparison of the Platinun } Hydrogen Thermometers in Liquid Air.\mdash ; The hydrogen thermometsr , employed for this purpose , has already been describedby one of us in conjunction with Drs. Senter and Jaquerod .
The method of [ vorking with and the degree of accuracy of the readings have been disoussed in detail , and it is only to state that the maximum errors never exceed at C. At the ice point the pressure of the gas in the thermometer was about 1000 mm. The platinum thermometer was held in a clamp so that the coil was almost in contact with the bulb of the platinum thermometer .
The liquid air surrounding thsm was contained in a cylindrical silvered vacuum-vesse120 cm .
in diameter and 5 cm .
deep , which , in the second and third set of experiments , was placed inside a somewhat larger silvered vacuum-vessel .
Some liquid air was poured into the space between the two vessels , and both the opening of the annular space and the mouth of the inner vessel were plugged with sheep 's wool .
With this arrangement the rate of rise of temperature as the air evaporated amounted to per minute .
And under these conditions it can safely be assumed that the temperature of neither thermometers differed from that of the bath by an eciable amount .
It is noteworthy that under such conditions liquid air appears to evaporate from the and does not undergo ular fractionation .
The following are the results of our experiments:\mdash ; * ' Phil. Trans , vol. 200 , 1902-3 .
Dr. M. W. Travers and Mr. A. G. C. Gwyer .
[ Jan. 11 , .
3322 ohms . .
003899 .
In this set of experiments the liquid air lvas contained in a single silvered vacuum vessel .
B. .
3319 ohms .
C. The second and third sets of experiments , in which the rate of rise of temperature was exceedingly slow , are probably the most accurate .
If we calculate the value of for the Centigrade temperature-190 for value of corresponding to and , we find 1900 On Scates of This difference is of the same order as that found by other investigators .
Compari.of th Solid Acid Alcohol.\mdash ; A mixtnre of solid carbouic lcid i ether or ectified methylated spirit frequently been mployed as a means of maintaining a constant temperature .
states that the temperature so long as the mixture has the oonsistencv of butte : but that as soon as on the surface , the tnre begrins to rise .
He the temperature as C. Our experime11ts to the conclusion that to obtain the best resnlts the solid onic acid should be free from , and should be added to the spirit\mdash ; for there is no advantage in using ether--till the mixture has the consiste11cy of fairly thin paste , and can be stirred a glass rod .
The stirring should be at short intervals .
VOL. LXXIV .
$ Dr. M. W. Travers and Mr. A. G. C. Gwyer .
The gas thermometer was the same as was used in the hquid air experiments .
As , however , the volume of the bulb was known only at C. , * the volume at-80 was oalculated ) means of a parabolic formula with two constants .
We have taken as onr values for the volume\mdash ; The rrangement of the thermometers and the vacuum-vessels was the same as in the liqnid air experiments , the only difference in the manner of carrying out being that the bath was stirred at intervals .
The following are the results:\mdash ; A. .
3319 ohms . .
003899 .
B. .
3319 ohms .
Though these results point to the conclusion that a mixture of solid carbonic acid and alcohol varies in temperature by less one degree over a period of more than 36 hours , it is only during the first few hours that it remains steady enough for this purpose .
Taking the first three observations of each set we obtain the resuIt\mdash ; Mean .
Mean T.H. scale . .
' Phil. rans , , vol. , p. 138 .
1905 .
] On the Normal Scalcs of Temp ) Taking the value for , obtained by comparing the thermoJneters in liquid air , we obtailt for the jtll .error of less than Mmnf the Freezing ) \mdash ; We have not succeeded in making comparisons of thermometers at temperatures other than and the temperatltre of liquid air .
The use of liquefied gases such as nitrous oxide ethylene in quantities up to 100 .
did ot give sufficiently temperatures , and larger quantities could not be procnred except prohibitive cost .
Attempts also made to compare the thermometers in ) aths of solids , such :as ethyl acetate , but withont success .
parison of Tran$ition Point ) ulphate.\mdash ; Richards and Wellshave shown that the transition point of sodium sulphate decahydrate to the anhydrolls salt may be considered as a standard temperatut.e in thermometry .
They measured this temperature on three Baudin thermometers and on one Tonnelot thermometer , and found , as the mean of results whioh differed themselves by , the value on the 1lormal scale .
For the purpose of measuring the tra1lsition temperature on the platinum scale , we prepared some pure sodium stllphate , dried tho crystals , and ground them to powdcr .
The powder was beaker in a water-bath at bout 4 C. till the mass became pasty , and was then transferred to the vcssel surrounding the which was enclosed in a tube paraffin oil ( p. 53 The vessel containing the sulphate was glass c.vlinder , 8 cm .
vidc cm .
deep , with a round bottom .
The cylinder was supported thermostat , and it was found that when the tempcratnre of the lattcr was kept at C. the temperature of the sodium snlphate rcm quite steady , and was not affected by Date . .
R. 7 .
Oct. 26 , , , The value of to the meaIl value of is it will be observed that this is with that obtained , 111 very close to the value at C. , viz. , .
Indeed , ppenrs t abolic formula may be applied to the calculation of and C. with a maximum error , probably not C. at about C. This result is mainly dependent 011 ' .
Phys. Chemie , ' 1898 , vol. 26 , p. 690 ; 1903 , vol. 43 , p. 465 .
Dr. M. W. Travers and Mr. A. G. C. Gwyer .
[ Jau .
11 , racy of our determination of the transition point of sodium sulphats 011 the platinum scale and that of Richards and Wells on the hydrogen scale .
If this resuIt is correct would appear that since the value for a gives for at C. , there must be at this temperature a difference of between the constant pressure air scale the constant volume hydrogen scale .
xrison of the at Higher \mdash ; In order to make certain that our thermometer behaved normally higher temperatures , we repeated the original experiments of Callendar and riffiths , obtained the same result .
Boiling Point of aphthalene , and on the .\mdash ; The apparatus employed in these experiments exactly similar to that described by Callendar and Griffiths .
In the case of the two organ ic lbstances the top of the glass Meyer tube closed by a cork , and a bent glass delivery tube served to condense the pour and conduct the liquid into a receiver .
The c.onstancy of the temperature while the substance distilled served as an indication of its purity .
The naphthalene was first dlised f absolute alcohol , WttS then bsequently redistilled before each experiment .
The benzophenonc ( Kahlbaum ) was dissolved in alcohol , water was dded to the hot solution till it appeared , and then ether was added till it ) ecanle clear .
On nding large crystals these were taken , dried in a desiccator , ttnd distilled .
In the sulphur experiments we sometimes employed pure sulphur , sometimes 11de snlphur , without any difference in ths result .
Naphthalene.\mdash ; Callendar and Griffiths taking for the of suIphm .
on the constnnt pressure thermometer , measured the boiling point by means of a platinum thermometer , and found the value .
Craftssured the boiling point by means of thermompter and ecently Jaquerod have deterIl the vapour pressure on the scale of a vohune mometer , filled with hydrogen at pressure of 500 mm. the ice .
Their results for the temperatures to pressure the neighl ) ourhood of 760 mm. may be expressed in the form .
1905 .
] the Platin Results .
It wiIl appear that our results , calculated on this basis , higher than those of Jaquerod ssmcr , and higher those of Callendar and Gl.iffiths .
The agreement the latter is , however , good to prove our point .
Benzophenone.\mdash ; Callendar Griffiths found the point to be while Jlerod and be expressed i11 the form .
The resnlt of the xperiment which we ont is follows\mdash ; R. .
Nov. 18 , 1904 The result calcnlated in this manner is igher t ] that of erod and grees exnctly with of Callendar Sulphur.\mdash ; According to and ( riffiths the boiling point of sulphur on the pressure air therlnometer is given by the expression .
Chappuis and Harker find the boiling } ) of the constant volume nitrogen scale the value Results .
Date .
As the of these results we may 'Brit .
Assoc 1899 , p. 24 Sir N. Lockyer and Mr. .
E. Baxandall .
[ Jan. 3 , or king Chappuis and Harker 's value for the boiling point , which higher , we obtain for onclusion.\mdash ; As might be expected , it is possible to apply the parabolic formula of Callendar and Griffiths to the recalculation of the differences .
between the platinum scale of temperature and the scale of the thermometer , though the range through which it is applicable , and the of constant , precludes the possibility of employing it except for interpolation .
A standard scale of temperature , based on Callendar 's three fixed points , usin standard wire , and taking the value of lvould obviously lead to absurd results at low temperatures ; and the converse may be said of our own observations .
To sum the matter we will tabulate the results that have been referred to in this paper .
Nulure of gas thermometer .
Observer .
Constant pressure air . . . .
Callendar and Griffiths Constant nitrogen Chappuis and Harker 154 volume nitrogcn standardised Harker. . . . . . . . . . . .
by constant pressure air ( 500 to ) Constant volume hydrogen to verse and Gwyel " " The Arc Spectrunl of Scandiunl and its Relation to Celestial Spectra By Sir ocKyElt , K.C.B. , , D. , Sc. D. , and .
E. ) .
Received January -Read uary 9 , 1905 .
Very little has been published regarding the spectrum of this rare element .
Thd records of Thalen , and Exner and Haschek , are the only ones given , the former observer confining his attention to the spark spectrum , whereas Exner and Haschek have recorded the lines under both arc ctnd spark conditions .
In the latter however , no lines are given in the region less refrangible than .
Rowland , in his " " Tables of Solar Wave-lengths , \ldquo ; certainly ascribes a small number of soJar lines to scandium , but , of course , no indication is there given as to the reIation of these lines to others which occur in the scandium spectrum , either in regard to number or intensity .
In connection with the work at Kensington on stellar and other celestial spectra , it httS been founcL that in some types of spectra scandium is conspicuously represented by some of its lines , in *(Ofversigt af Kongl .
Vete Akad .
' vol. 38 , No. 6 , ' -Tabellen fur Spektralanalytische ungen auf Grun der Ultravioletten Funkenspektren und Bogenspektren der Element , ' Leipzig und Wien , -Deuticke , 1902 .

