A N E W FORM O F AIR THERMOMETER FOR TECHNICAL PURPOSES.
A comparison of several ordinary thermometers in use in a varnish factory showed, as was anticipated, considerable differences in reading under similar circumstances, these differences amounting to as much as 50" E'. a t a temperature of 500" F. or thereabout. A part of these differences could be ascribed to age, one of the instruments being about fifteen years old ; but only a small part of t h e differences were due to any such cause. Inquiry of parties who make a business of correcting thermometers disclosed t h e fact that very little had been done in t h e way of correcting thermometers for high temperatures, and t h e conclusioiis finally reached were as follows : It appears to be customary to assume the expansion of mercury to be uniform; but this is said not to be t h e case when we approach i t s boiling point, and I have not found any statement as to its expansion a t high temperatures under pressure, as in the nitrogen filled inst ramen ts. It is customary to assume the expansion of glass to be uniform ; b u t this is not true when we a p p ~ o a c hthe softening point, and in any givcii instrument i t is impossible to tell how much of a n error is thus occasioned. In graduating a thermometer i t is customary to assume t h a t all t h a t part of t h e instrument which a t the time is full of mercury is immersed i n the fluid which is under examiliation ; and t h e error from this cause is stated by an -.xpert to be probably BP much as SO" F. a t temperatures approaching t h e boiling point of mercury. T h e improved method for graduating instruments for use a t high temperature appears to be by the use of substances of high 2-77
boiling point, such as naphthaiine and bcnzophenotie. l < n t different observers find different boiling points, eve11 for a simple substance like mercury, and to get cven approximate accuracy t h e experimenter would be obliged to determine with a n :tir thermometer t h e boiling point of each substance he used, under t h e conditions of iisc. Finctlly, t h e conclusion appears to bc inevitable that tlie thermometer must be standardized under conditions essentially or absolutely similar to those under which i t is to be used. T h e colditions in t h e particular case in which I am interested are as follows : T h e thermometer titbc is ;tbout t h e e feet long, the graduation being confined to the upper one-third, and is inclosed in a tubular case, which termiiiates at t'lie bottom in a steel cylinder which snrronnds tlie b u l b and is liartly filled with mercury. !Phis instrument is p u t in a copper kettle about two feet deep and has about a foot of tho lowdr part of the instrument immersed i l l hot oil. S o w an>- rcading based on the ordinary graduation will under thcse circumstances bu so erroneous as to be of 110 absolute value,, and i n f w t , varnish ni:ikers d o not go by degrees: but p u t arbitrary marks 011 their ilietruments id heat 111) to the mark, these marks not being a t tlic same scale points of different therrnomcters. T desired to m:tke a thermometer wliicli w9ultl give a reading within 5' F. of absolute accuracy, aud the means I employ are a s follows :
A XEIV POHJI OF A I R THER3IOJIETER.
%-
-
279
--- -Y
D E S C R I P T I O N OF THE D I A G R A M .
A - - - -Bulb. B - - - -Safety chamber for air and mercury. C - - - -Mercury cylinder. D - - - -Air pressure gauge. E - - - - Capillary tube. F- - - - Piston with screw to force mercury from C. x y--Horizontal line showing level of mercury in capillary tube and gauge glass. T o avoid the irregular expansion of glass heated to near its softening point 1 use a bulb of metal, iron being selected; all the experiments which hare been made tend t o show t h a t t h e rate of
280
A S E W FOR31 O F A I R THERXOMETElL
expansion of iron a t temperatures not exceeding 800" F. is very uniform, and on general principles we should expect this to be the case, since such temperatures are far from its softening point. A bulb of cast iron containing about 160 C.C. is fitttd 11y :LHasbestosp a r k d joint t o a ciipil1:q glass tube : the capacity of the hull) is of n o consequence except t h a t i t s h o u l d be large enough to make t h e capacity of the capilhry tube insignific:int. 'l'his glass tube is of any convenient lcngtli, say 3;; feet, arid after bending downward is connected with ;t ulianiler to prevciit t h e escape of air, iis shown in the accompanying diagram. T h i s safety chamber is iiormiilly filled with mercury, which is forced i i i t o it liy :L piston which is screwed into a cylinder filled wit,ii mercury w li i c I1 tom m u n i cii t e s m i t t 1 t li i s safety c 11ani bc: r, ;I I i (1 1)y t 11 is means the thermonieter is made a constant volume one. T h i s mercury also connects with a suitablc 1)rcssure gauge ; the form whirh I iise is it closed tube coiitaining a i r Jvliicli is connectod by a rnbber tube witli tho sRfety cliamber. Owing t o the flexibility of the connection the gnugc tube can always be placed so t h a t t h e mercury in i t sliall l ~ ea t t h e lcvc~l of the merc u r y i n t h e capillary t u b e iind thus m y correctioii on that accotint is avoided. and the whole apparatus ljeilig closed 110 correctioii for barometric pressure is needed. Of course an open mcrcnry giiuge tub(. c~oiild be used just iis well if clesircd IYIiiit occurs when the l ~ i ~ iisb 1icAted is this : 'l'lic expaiisioii of tho air i s exactly countt~rbalancedb y the operator who s(:rows down ti!c piston of the r n e x u r y cylinder so as to keep tlie mercury a t t h e sanie ]mint in tlie capillary tube : the only effect t,lien of the Iicat is to increase the pressure on t h e mercury. T h e mercury transmits t h e pressure to t h e air in the gauge glass, which is compressed, and from the amount of this compression the tempcrnture is calculated. T h e calculation was made in t h e following manner : T h e expansion of air is 1
1
of its 0" C. volume for each degree C'. i t is heated; 1
or --'its 32' F. volume for each degree E'. it is h e a t e c l . ~ = 491.4 .002033, and 1 volume air a t 32" F. becomes 1.07733 ~ o l a. t 70" F. Since 70" 1:. is the lowest ordinary temperature in the places
A NEW FORM O F AIR 'THERMOMETER.
281
where I am likely t o use this instrument, for reasons which are obvious, I have selected 70" F. as my normal temperature. Other observers of conrse would calculate other tables. Now if we assume 70" F. as normal, what will be t h e expansion of a given volume of a i r ? T h e given volume will fill the space which would be occupied by LOT733 volumes if the temperature were 32" and 100000 therefore consists of cc- 9283 of t h e amount of 32" air 1 0 1i33-' which it would take to fill the space: and as its expansion is calculated on 32" air, it will expand ,9282 of .002035 = .00188 of its own 70" F. volume for each degree F. i t is raised in ternperat 11re. If air a t 70" tends to expand .OOlSS. . , of its volume when heated 1" and is prevented from doing so by being enclosed in a rigid vessel, t h e pressui'e is equivalent to that produced by pumping in .001888 of t h e original volume and keeping the temperature a t 70". I n t h a t case we should have 1.0018SS vol. condensed to 1.0000 vol.; and t h e original 1 rQlumewould be condensed to .998113 as found by the proportion 1.0018S8 : 1 : : 1 : .998113 ; a n d so on. T h u s a rise of 10" F., t h a t is, to 80' F., would produce a condensation found by this proportion : 1.01888 : 1 : : 1 : .9314 and t h e space occupied by the condensing agent, which is t h e mercury in t h e gauge glass, is .OlSG of the normal volume of the gauge, for this interval of 10" F. To get the volume a t each successive int e r v d of 10" F., multiply .01SS., , by one, two, three, and so on, and add the product to unity for t h e first member of the proportion, whence we get a table, which does not, however, allow for t h e expansion of t h e iron bulb, nor for a change of temperature in t h e air of t h e gauge glass. T h e cubical expansion of cast iron is .000033G for 1" C. If t h e shell be heated t o 800" F., from '70" F., which is equal to 405.6' C., it will expand .000033G~405.G=.01363 of t h e original volume of t h e bulb. A t 800" F. the volume of t h e air in t h e gauge is .4203 o€ its nornial volume ; a n d .4203>(.01363 = .00573, this latter fraction being the part of the normal volume of t h e gauge which t h e air in i t will expand on account of t h e increase of t h e size of t h e iron bulb a t 500" F. T h i s makes a difference in reading such
....
as would be caused by a diftcrcnce of about l L S L in temperature. Actually the volume of tlic air i l l thc gauge is .4303+.0057:1= .42G03 of its 11orn1al volume, wlieii t l i c s bulb is heated to SOO" 1.; In this manner a corrected table maj- i)c: derived f r o m t h e t:ible alrcady c:tilcul:ited, : ~ n dthis corrcctecl table w i l l show t h e absolute tcmperaturo of tlie tiir iii til(: bulb, w i i c i i t i i t , air i i i the gauge glass is a t ;Oc 1:. .it .!OO'' 1:. the volume of tlic? air i n the gaugc glass i f a t TO", is .Wl; of its riormirl volumc. S o w if tile original volume were raised t o SO" F. it woultl liavc tlic: voluinc 1.01888, aiid if this volume were subjected to sufficient pressure to comprcm it to .8CJ4'; of its bulk it would occupy .S047X 1.01S8S=.S;(j?. 'I'his exc:ceds t h e 70" volume by . O ' X 5 , silicic .8;(i?-.SO1; =.O;lj: a n d thcb difference f o r 10" in t h e bulb a t tiiis temperature ( W O " F.) is .011!1 : ;tiid
.o; lr, -iij.OI
.0 11!) -
: hencc
t ~ i icS1ev:itioii
or' L O
iii
the gauge
air temperature makes ;L difference of 10 x (i.01= ( j O . 1 ' in the reading and this amount shoultl be added to the observcd temperature under sucli coiiditioiis : and this ratio of ti.01 : 1 is true not only of tlie temperature of tell degrees in cxcess of io', ill the gauge glass, b u t also of :dl ternperrttnrcs i l l excess of wheii the bulb temperature is ?ooc F. 111this wag a tit& of corrections may be calculated for t h e various bulb temperatures to be observed. The ratio increases ;is the temperature rises. The air gauge of t h e instrument should contain it sinal1 thermometer, and a sliding scale can be arranged if desired.
