]\gt ; tisation .
On the Changes of Thermoelectric Power produced by sation , and their Relation to metic Stxains By BIDWELL , JM .
A. , Sc. D. , F.R.S. Received April ll , \mdash ; Read April , 1904 .
It is well known that ticn generally produces a change both in the thermoelectric quality of a , and also in its linear dimensions .
In an article published October , I directed attention to the remarkable tive c which appeared in several cases to exist veen the two sses of , and the xperiments which form the of the csent paper undertaken with the view of ting their apparent tion .
Some of the results were of an u1lexpected .
The accepted statements arding certain thermoelectric effects to which this connection special importance was ttached , to be not generally true , and , far my owll go , altogether erroneous .
Several experimenters to misled by regarding as unmagnetised a piece of metal which cither ined permanent magnetism or was ith the piece ) jccted to magnetisation .
What they fact the power of magnetised with respect to metal , } ) of more strongly magnetised with resl ) to strongly netised ; and the effects may be , as will direc.tly opposite in the cases .
Probal ) also mistakes have risc from the ssumption that for a lengthened period ths electromove forces in the vcnt uo changes other than those due to etisation .
It is rdly p to keep the tempel.atures thronghout so to void gradual changes of electromotive force consi than those which it is desired measnre .
own utions were taken gainst bolh these of osel v tion the metal was dem.agnetised ] the nometer reading for no noted .
Although , as mentio/ led , some of the chief grounds which formed asis of my ljecture ffl to the electric and the effects have disappea ] cnce is evertheless f , that for iroll and nickcl thore is a relation , and orcly qitative .
-As trds important thel .
is ined inconsistency the behaviour of the two ) the coincidejlccs ) in both cases are too numcrous too varied to be the resnlt of nccident .
Iron.\mdash ; For iron , wheIl aance is for the pnrely mechanical * .
Brit vol. 30 , p. 449 , rticl \ldquo ; VOL. LXXIII .
2 Dr. S. Bidwell .
On the Changes of compression due to magnetisation , the change of the1moelectric power appears to be proportional to the change of length .
Change of thermo .
electric power expressed in microvolts is nearly ( and perhaps under perfect experimental conditions would be exactly ) numerically equal to the " " corrected\ldquo ; change of lsngth in ten-millionths multiplied by a factor which is constant for the same specimen in the same physical condition , but differs for different specimens , and for different physical conditions of the same specimen .
For my sample of pure iron when in a force state , the factor giving the best agreement is ; for a sample of good commercial iron it is and for the pure iron stretched by a load of 16201rilogrammes per sq .
cm .
its value is .
The curves expressing the relations of thermoelectric power and of change of length to magnetising force are not indeed exactly coincident , as may be seen by reference to and 5 ; but since identical specimens of the metal were not used in the two sefs of experiments , while the condibions were necessalily somewhat different , the divergencies cannot but be regarded as very small ; somstimes , indeed , they hardly exceed the limits of experimental error .
The thermoelectric and the elongation curves for iron appear to be similarly influenced by the physical condition of the .
I have shown that the elongation curve is lower for annealed than for mannealed iron , and that silc stress also lowers the curve ; the same is the case with the of thermoelectric power .
The strength of the magnetic field at which , as indicated by the corrected curve , there would be no changc of under tensile stress , appears to be just the same as that at which the thermoelectric force becomes zero ; when this strength is exceeded , retraction of elongation , and a simultaneous reversal occurs in the direction of the thermoelectric force .
Unlike previous experimenters I find that when the iron is free from tensile stress the direction of the thermoelectric force is never reversed by magtletisation , even in fields up to 1600 C.G.S. units ; neither does the of change of length when " " corrected\ldquo ; for mechanical stress ever cross the horizontal axis .
Nickel.\mdash ; Partly on account of the smaller magnetic snsceptibility of nickel , and partly in consequence of the relatively changes of length which that metal mdergoes when longitudinally magnetised , the corl.ection for mechallical stress is almost negligible , averaglng than 3 per ccnt .
for fields up to 1200 .
Here , too , the forms of the curves for change of length and change of thermoelectric power in relation to are strikingly alike ( see fig. 7 ) , the spondence being even closer than in the case of iron .
For a specimen of pure nickel the of rmoelectric powsr in microvoIts , due to is a disputed point whether there actually is any such ubject is discussed later .
'Roy .
Soc. Proc vol. 55 , p. 228 , ; vol. 47 , p. 469 , 1890 .
Thermo ( jlectric Power produced by Nagnetisation .
tion in any field uP to 1600 ( the strongest reached ) , was found to be about equal numerica.lly to the retraction in ten-millionths multiplied by .
The curves for two pieces of an impure nickei also exhibit general similarity of form , but since the dimensional ratios ( ratio of length to cross section ) of the two pieces are very different , the curves are not strictly .
The changes of thermoelectric forcs are , like the changes of length , much greater for nickel than for iron .
Tensile stress produces , as in iron , corresponding variations in the two classes of curves .
The effect of tension upon the magnetic retraction of nickel is , as I have shown in a former paper , not so simple as in the case of iron .
In weak fields the contraction is diminished by tension ; in fields of more than nbout the contraction is increased by tensile stress up to a certain critica ] of the stress depending npon the strength of the field , and diminished by greater tension .
Thus it happens that the retraction ctlrves for a wlrs loaded with two different weights may cross othel ' .
In one of my published iments the retraction curves for a nickel wire carrying of 420 and of 980 kilogrammes per sq .
crosfied when reached 220 .
With nearly the same loads , the two of thermoelectric force also crossed , though in a weaker field , ) eing only 150 .
lVire of the same quality was used in both expel.iments , but in the first it was hard , while in the second it was anncaled , which ] may or may not be the reason of the difference .
In any case , it is interesting to find this complex and unexpected phenome1lo qualitatively reproduced by the thermoelectric curves .
The anomaly above referred to consists in the fact that the direction of the thermoelectric duc to etisation is the same for nickel as for iron , whereas is oppositcly in the two motals , iron being extended , nickcl contracted .
Cobalt.\mdash ; For cobalt I have not succeeded infinding amy relation between the thermoclectric and nges magnetisation .
The thermoelectric cve somewhat reselubles that for nickel , but it is much lower .
The strain curve is opposite in to that of iron ; in fields the nletal contrncts , in strong fields it is gated .
When the direction of Cllrrent metals , A and , was fionl A to hrongh tction , it formerly the cubtom ) that A was positis to I3 to ; bismuth , for nplc , to ) positive to antimony .
Of } custom to exteut reversed , and the resenlt in unsettlod condition .
Following what I to be the best crll { mthorities , I in * ' Roy . .
Proc vol. 47 , p. 174 , 1890 .
Dr. S. Bidwell .
On the Changes of this paper speak of a metal A as being positivs to when the thermorent passes from A to the col junctio1l , or from to through the hot .
The thermoelectrio power of bismuth with respect to lead is thus negative , while that of antimony is positive .
us The results of my owll experiments compel me to dissent from some of the conclusions which have been reached by other workers .
The discrepancies noticed , I think , in most cases due to the fact already referred to , that the thermoelectric behaviour of weakly magnetised with respect to strongly magnetised iron and impure nickel may be very different from that of the unmagnetised with respect to the magnetised metal .
The earliest tions of the effects of magnetisation upon thermoeIectric power are those of Professor W. Thomson ( Lord Kelvin ) , in 1856nnounced that magnetisation rendered iron and steel positive to the unmagnetised metals ( the thermo-current passing from unmagnetised to gnetised hrough hot ) , while in nickel the opposite effect was produced .
The Iatter statement , though it has never , I believe , been disputed , is doubt erroneous .
Iron , steel , and nickel , when free from mechanical stress , all become more positive when magnetised .
The effects of tension and magnetisation upon the thermoelectric quality of iron have been investigated by Ewing , who foumd that " " the of diminishes the general thermoelectric effect of magnetisation , and ally reverses it when the load is great Although this happens to be a oorrect statement of the facts , is , I venture to think , exceedingly doubtful whether a genuine reversal was actually attaiued in a field of only 17 units , the strongest ) lied by , even with a on the wire of 4000 kilogrammes per sq .
cm .
From the description of the apparatus ( p. 368 ) it appears that the piece of wire regarded unmagnetised was co1ltinuous lvith the magnetised portion , the hot junction separating the magnetised the gnetised iron outside the magnetising coil .
Chassagny the first to announce that the increase of thermoelectric power dus to magnetisation reaches a maximum in a moderate field and dimirlishes in stronger fields .
In his experiment the electromotive force was grcatest in a field of 55 units , falling to about half its maximum in a field of 200 .
In a similar experiment by HoulIevigueS a maximum was indicated at , and at the increase of thermoelectric force due Lecture , ' Phil. Trans 1856 , p. 722 .
'Phil .
' 1886 , p. 361 . .
' C. vol. 11 p. 977 , 1893 .
'Journ .
de sique , ' vol. 5 , p. 63 , 1896 .
Thermoelectric Power tion .
to magnetisation had rently become zero .
His paper record of afi actual reversal of the force in fields , though the of such reversal is snggested .
experiments , which vere made with several different specim of , show no approach to zero in fields five times as strong as mcntioned .
As regards the behaviour of stcel , Houllevigue 's results are at variancs both with those of W. Thomson and with my own .
Mr. E. Rhoads* has given an account of two thermoelectric experiments , one of which was with iron , the other with nickel , his work having been undertaken " " in looking for some property of iron that would vary with the ntagnetisation in the same way as the length After referring to the experiments of Chassagny , and to those of Houllevigue " " who found that the diminution [ of thermoelectric force ] continues in higher fields , and thfit a roversal ctually takes place he thut " " snggests change of length , and made me wish to work out the cyclic curve for comparison lvith it The cyclic curve which he obtained hysteresis , and bears a gsneral resemblance to one given for of length , the latter is not carried beyond .
The thermoelectric curve for iron is represented as crossing of at , continuing to fall in an almost ight line up to , the strongest field applied .
Rhoads considers that il agrees with the curve for change of length when the latter is corrected for stress\ldquo ; ( which is given as ) being apparcntly of the demonstrated by More , and afterwards by KliugGIlI } , that the corrected curve never crosses the horizontal axis .
The thermoelectric ourve nickel is shown in Rhoads 's diagram as , likc the lrve for change of length , below the axis ( in with Thomson 's while for iron , both the therrlloelcctric and the corrected lsngth curves are above the axis .
' The propcrtics he , " " ttlrn out to be related in the opposite sense in the two \ldquo ; As regarck form , the nickel curves for change of thermoelectric power nd of length are of the same chnracter , though , as is marked , they are not strictly , being made with very speeimens .
No mention is made of Maxwell 's stress in tion to nickel .
to from .
Before the xperimonts it is to discuss briefly the controverted questioI of comprcsve stl such important npoll the results .
Iu a ) on the changes of * Revicw , ' vol. p. 321 .
) later than my article to .
My ) forc 1 ' heard of Mr. 's work .
atlS .
' vol. 179 , p. 216 , 1888 .
Dr. S. Bidwell .
On the Chang of dimensions due co magnetisation , pubIished in 1888 , I wrote : ' On{ ; of the influences tending to produce retraction must certainly be of a purely mechanical nature .
Suppose a uniformly magnetised rod to be transversely divided through the middle .
The two halves , if placed end to end , will be held together by their mutual attraction , pressing against each other with a certain force per unit of area , which can be measured by the weight necessary to tear one from the other .
The sams pressure will exist between any two portions of the rod separated by any possible cross section , and a certain longitudinal contraction of the rod will be the consequence .
If , now , the rod , having been first demagnetised , be placed in a vertical position upon a fixed base , and loaded at the upper end with a weight equal to the greatest it could support when magnetised , it will undergo the same contraction as before , the [ compressive ] stresses being equal in the two cases It pointed out that in the latber case the contraction is expressed as a fraction of the original length being Young 's modulus , and the load , both in grammes weight per centimetre .
Assuming the magnetisation to be such that the divided rod can just support a weight of grammes per square centimetre , it is inferred that the contraction due to the mechanical effect of magnetisation would again be , and it is shown that this accounts for a part only of the observed change of length .
In an earlier paper it was calculated that , the weight per squnre centimetre supported in the field , was equal to , I bein the netisation , and the intensity of gravity .
This expression ( which is equivalent to ) , applicable when the magnetic metal is magnetised longitudinally and by an exter/ lal field , as was approximately the oase in the experimen ts with which the present paper is concerned , where the wires were placed in the axis of an independent magnetising coil .
For the special case in which each half of the divided rod is surrounded by a separate magnetising coil pped tightly around it , another term , , must be added for the mutnal action of the two coils , and we shall have Also for a permanent ring-magnet , in which there is no magnetic force It should be noticed that since , except in very slrong fields , is negligible in comparison with , the force in the first case considered * ' Roy .
Soc. Proc , p. 486 , 1886 .
Th Magnetis ption .
generally without sensible error be taken as , instead of or In the year 1895 it was pointed out by Iore , who was apparently ignorant of what I had written on the snbjoct , though he was certainly familiar with most of my work , that the change of length attending magnetisation must be due to several causes , amonlg which the mechanical stresses created in the rod .
" " The firsl of these mechanical stresses is the tractive force of the magnet , and is measured by \ldquo ; This force , he remarks , tencls lways to the rod , and for high intensities becomes one of the most important factors the observed changes of length .
In plotting his curves ( in which abscissae are I instead of H ) he makes correction for the contraction due to the force .
This correction is obtained from the formula is the modulus of elasticity .
The eifect of this correction is to mnke the elongation much for a given intensity .
The maximum value of the elongation is more than twice as great as the observed maximum , and the greatest tensity employed , 1300 C.G.S. , procduces an elongation and not a contraction as observed The publication of More 's paper led to discussion which several well known physicists took ; but ths views essed 1 } by no means concordant , and I believe that the prcsent time it is not agreed whether there is in fact any such mechanical stress ; whether , supposing one to exist , it is compressive or tensile , whether it is " " Maxwell 's stress\ldquo ; or some other .
The compressive stress which I contemplated was , of course , quite unconnected lvith of stress in the electro-magnetic field , and the expression enlployed for its value was based upon principles which were well long before the date of work .
S Several papers have been more recc1ltly published in which the * ' Phil. Mag vol. 29 , p. 440 , rence due to the ] is , when cent. ; when per cont .
; and when per cent. ' Phil. Mug vol. 40 , 1 ) .
345 , 1895 .
' Nature , ' vol. 53 , pp. 269 , 316 , 365 , 462 , 633 .
SThs gnetic force side rrow in a dinally magnetised bar is npposing one of } .
to be fixed , the force act.ing upon the of othel porbion is loss } : , the due to the itself ; thus ) cve force .
umit .
The stress betwcen any two portions of bar , di vide by an inary tnverse plane , is inecl by the in , whatever their ph3 bical l be , to elasticiby of the is due .
Klingenborg , tock Univ. Thesis , ' , 1897 ; l ; rackett , .
Rev vol. 5 , p. , lS97 ; Rhoads , 420 Dr. S. Bidwell .
On the es of writers sist upon the nec .Qsity of correcting curves of magnetic for mecharlical stress .
Brackett attributes the recognition of necessity to Rowland , who , he says , defended his position as follows:\mdash ; The compressive force which will tend to close up a very thin air gap in a divided magnet ust exist in any magnet , for according to all our ideas of matf er thsre no real difference in the case where the air gap exists and where it does not , because we must still consider the between the moleoules .
I venture to think that the results of my thermoelectric experiments with iron nickel afford strong evidence of the reality of the mechanical stress in question .
Before any comparison is possible between the two phenomena of change of length and change of thermoelectric power , considered as due to the molecular effects of magnetisation , it is clear that the effect of any extraneous mechanical action tending to alter the length of ths metal must be eliminated .
The stress under discussion , if it exists , is our present purpose no less an extraneous one than if were produced by loading the metal with a weight .
Conjoctm.ing that there might be a relation between the thermoeleCtric and the strain efiects of magnetisation , I plotted curves for the two phenomena side by side ( see curves ( D ) and ( b ) , fig. 4 ) ; but although these curves bear a superficial resemblance to each other , it is clear that the phenomena to which they relate are by no means in correspondencs .
In particular , one curve cuts the axis at an early stage , indicating a reversal of sign , while the other appears to become asymptotic .
Curve ( d ) hich is drawn to the sams scale as indicates , in tenmillionths of ) , the correction to be made in respect of the hypothetical compression due to mechanical reSS , its ordinates are tional to being the greatest weight which ( as found experlmentally ) could be lifted when the iron was ) etised by the corresponding field , being Young 's modulus .
Adding together curves and ( d ) , we get the " " corrected\ldquo ; cve ( E ) , which , as will be seen , coincides with the thermoelectric curve ( D ) as nearly as could be expected uIlder the conditions of the experiments .
Still more striking is the case of the wire under tension , curyes ( F ) and ( f ) fig. 5 .
Here there is not even a superficial resemblance ; the curve for change of length is an almost straight line inclined to ths axis and lying entirely below it , while the thermoelectric curve begins above the txis and crosses it at .
By making the same addition for stress as in the former case , curve ( d ) being again employed for the purpose , we ) tain a corrected curve of length , indicated by the crosses on curve ( F ) , which , when the ordinates ars plotted to a scale , closely follows the thermoelecbric curve ( F ) , and would , if it vere continued a little farther , evidently meet the axis at just the same point .
All the curves for c. hang of referred to in this have appeared in There moelectr.icPo}produced So far the presumption in favour of the reality of the compressive appears to be fairly strong , but it is , I think , greatly strengthened the results obtained for nickel .
In fig. 7 and ) are curves of change of thermoelectric power and of change of length for this metal .
There is obviously a strong likeness between the two , though they lie on opposite sides of the axis .
Curve is derived from ( k ) simply by inverting the latter and plotting the ordinates to a slightly different scale .
The inverted curve is seen to be throughout its whole length almost a counterpart of the thermoelectric curve and this although it not been " " corrected\ldquo ; for compressive stress .
YVhy then should the correction which is indispensable for iron be unnecessary for nickel The answer is that while for iron the calculated correction for mechanical stress is relatively very considerable , for nickel it turns out to be very small , so as to be negligible .
The changes of length indicated by the compression curve for iron , ( d ) , fig. 4 , are generally much greater those ) the curve ( b ) , to which the correction is to be applied ; and the two curves have Yery different forms .
On the other hand the compression curve for nickel , fig. 7 , is an almost straight line making a very small the axis of , the changes of length which it to not more than 2 or 3 per cent. of changes exhibited ) the uncorrected curve for correspo1lding fields .
Both curves , moreover , rise gradually from the origin to } highest points .
Thus it that the lcorrected and the corrected curves for nickel , if referred respectively to of ordinates so chosen that the two curves may be of the same height , are sensibly identical .
Curye ) may be regarded either as the uKlcorrected or as the eotsd c , according as the ordinates are to the given in the right-hand margin to a different scale in which each unit is very slightly increased .
Thus the absence of any need for the correction in the of nickel where prior.i it ought not to be required , tends to shcw that the success of its application in the case of iron is ) a more accident , and that the oompressive stress is consequently a Methods of cperiment .
Two of experiment were employed .
In the one the thermoelectric force of the magnetisable in conjunction with copper was opposed by a similar couple which the first metal was .
in the other it was force derived from battery .
The arrangement adopted for.thod is shown diagl.ammatically in former communicatiolJs to ) Royal 8ociety .
It is connection hat some of the experiments not arried f Dr. S. Bidwell .
On the Changes of fig. 1 , which , for the sake of clearness , is not.drawn to scale .
compound rod made by together snd to end five wires in the order ( 1 ) c.opper , ( 2 ) iron ( or other magnetic ( 3 ) copper , ( 4 ) iron , ( 5 ) copper .
The ]ength of each iron wire cm .
, and that of the copper wire between them 12 cm .
The middle and end portions of the rod are surrounded by three separate brass tubes , in the sides of which are inlet pipes , , and outlet pipes , , for the admission and egress of water and steam .
The rod is held axially in the tubes by well-tallowed corks , through which it passes , each of the four copper-iron junctions being exactly opposite an inlet .
Cold water enters at and and emerges at and ; steam from a small boiler is admitted at and , escaping at K. It was ascertained experimentally that with the low resistances and small omotive forces employed water was a good insulator .
The iron wire is magnetised by the coil PP .
The field at due to the coil is less than a hundrsdth parG of that in the interior of the coil , and is regarded as negligible ; but if the apparatus were to be reconstructed , I should , with my present experience , consider it desirable to increase the distance between and N. The two ends of the compound rod AB were joined to a galvanometer of low resistance .
Nefhod II is diagrammatically illustrated in fig. 2 .
consists of a wire or strip of the gnetic metal soldered between two copper wires ; it passes through two short and wide brass tubes provided lvith corks at their ends .
Inlet and outlet tubes are fitted to the corks , as shown in the diagram , Cadmittir } cold wate ] ' and steam .
Cold vater also flows through , to equalise the temperatnres at and B. The electromotive force of the thel.mo-couple is opposed by the potentiometer arrangement shown .
is a dry cell having an electromotive forcs of volt .
The low resistance was either 10 or 2 ohms .
the high resistance ranged from some hundreds to some thousands of ohms in the differcnt experiments .
Before an experimsnt Power produced by was begun , was adjusted until the galvanometer showed no deflection when the circuit was closed , though a small permanent current was really of no consequence .
The deflection ( or increase of leileetion ) , which oecnrred when was etised thus entirely due to the change of thermoelectric force by magnetisation .
The aperiodie galvanometer , which } placed at a distance of metres from the magnetising coil , i@ of the yned by dand IsIather ; the resistance of its moying coil together with the , suspension is ohm .
In most of the the distance between the mirror and the scale was 12 feet , when the deflection for 1 nuicroampers was found to be 54 scale divisions of inch cm Two magnetising coils is .
in length and contains 876 turns of No. 18 , the eld near the middle of its interior due to 1 being 92 units .
The leilgth of the other is cm .
, and the number of turns of wire ] produces field of 174 units in its interior .
The metising c was derived from a battery of 27 storage cells was measured by a moving-coil ammeter , reading from 0\mdash ; 15 amperes by tenths .
The ammeter was calibrated by comparison with tangent nometer , and the readings were found to be sufficiently accurate for the purpose in view .
Cur } of more than 15 mperes were by the engine-room ammeter ; but on account of their excessive ting effect such strong currents were used in only a few of the imvnts .
Iron.\mdash ; The results of experiments with two different mples of iron and one of steel re by cves ( A ) , ( B ) and ( C ) in fig. 3 , where the ordinates are proportional to ths temporary increase of Dr. S. Bidwell .
On the Changes of thermoelectric force produced by the application of the external field H. Curve ( A ) relates to a specimen of iron mm. in diameter , snpplied by Messrs. J. J. Griffin and Sons as pure .
of the experiment are given as an example of the manner of working:\mdash ; Method II .
ohms .
Coil No. ampe Temperaturs of cold water , C. Rcsistance of meter ohms .
leading wires potential coil thermo-rod , Total resistance in circuit , , Electromotive force of 1 microvolt gives microamperes scalc divisions , or deflection in scale divisions microvolts .
Before every observation the iron was demagnetised by 's method of reversals , alternating currents dscreasing to zero eing switched into the magnetising coil .
The difference of the temperaturss of the waCer at and of the steam at was ; assuming that the actual temperatures of the junctions were respeclively above and below those of their surroundings , ths change of thermoelectric power for a giyen field Thermoelectric Power produced by .
425 found by dividing ths change of microvolts by 85 .
In all experiments the temperature of the cold water differed little , the mean temperature being therefore Curve ( B ) was given by a pisce of commorcial iron , which was hard , springy , and easily broken .
Changes of thernloslectric iorce , like changes of length , are greater for such a specimen than for one which is purer and softer .
Curve ( C ) shows the results for a piece of steel knitting needle in the in which it ought .
The experimsnt was repeated with the lvire in a glass hard state , but the curve was almost The dotted , copied from a former paper , shows for comparison the changes of lsngth sxhibited by roIl wire in fields up to 1500 .
The main point of difl'erence is that whereas the curvs for change of length crosses the axis of , indicating a reversal of the sign of the phenomenon in moderate fields , the other does not .
Two other thermoelectric xporiments , made with difl.erent pieces of iron , gave results of a like character .
Even lvith ths high field of 1600 , which greatly exceeds any before used in similar work , there is no indication that a reversal of sign would ever be ched .
The curves in fig. 4 to show the effect of ' ' correction " " for mechanical stress .
The pieces of wire used in the three experiments concerned were all cut for the hank , and imperfectly annealed by ting in a Bunsen Cnrve ( D ) is the thermoelectric for the sample , and the curve of elongation and retraction .
Cve ( c ) shows ths relation of lifting powsr to field , the ordinates being referred to the scale grammes weight on the square centimetre for a given value of , as shown in , and Young 's modulus in grammes weight per square centimetrs , the mechanical contraction in ten-millionths .
The value of is taken as .
( It might without sensibly affecting the form of the curve have been or , and its actual value no doubt comes within these limits .
) Hence , , or the mechanical contraction in ten-millionths is 1lumerically equal to the ammes weight supported divided by 200 ; ordinates of curve ( d ) are therefore simply .
In the following table values of are obtained by measuring ordinates of curve ; elongations and retractions ( E ) are similarly derived from the published original of curve Ordinates of the thres curves , ) , ( d ) , and ( E ) are referred to the scale of ten-millionths given in the middle of the diagram .
For the sake of easy comparison this scale was so chosen that the heights of the thermoelectric and the corrected elongation curves ( D ) and ( E ) are about equal .
As before remarked , the close agreement between the Thermoelectric pro agnctisatior curves can hardly be the result of accident , the less so the same similariby is manifested under different conditions .
Tension has the of loweriug the elongabion curve of iron , the maximum becoming smallet ' and contraction an earlier stage of the magnebisation .
If the tension is great there is no preliminary elongation at all , and the to contract wibh the smallest magnetising forces .
Curves and ( f ) , fig. 5 , show tho Dr. S. Bidwell .
On the Chcmges of [ Apr. 11 , nges of ngth undergone by an iron wire when loaded wibh weights of 351 and 1600 kilogrammes per sq .
cm .
, the curve for the first-named load bsing practically the same as for an unloaded wire .
Curves ( F ) and ( G ) indicate the changes of thermoelectric force with field for an iron wire , firsl when loaded with 1620 kilogrammes per sq .
cm .
, and afterwards without load , the points of observation being marked by dots .
The crosses on ( G ) and the course of the curves ( e ) and ( f ) after correction for mechanical stress ; the correspondence is so close that it was impracticable to draw the corrected curves separatel. .
The vertical scale of the corrected curves is not the same in the two cases ; that for the upper one is given in the right-hand margin , and that for the lower in the middle of the .
Data for ths consbruction of these curves were obtained in the same manner as before , and are given in the annexed table:\mdash ; is to bs noted that the three sets of experiments\mdash ; for change of thermoelectric force , of length , and lifbing power\mdash ; were made with three different samples of iron wire .
The effect of the changes of length and of thermoforce is illustrated in fig. 6 .
Curve ( g ) shows the elongation of a in the state in which it was bought , while ( h ) indicates its viour after it had been carefully annealed .
This operation was performed by enclosing the wire in iron tube which was placed in a hot fire and allowed to cool gradually as the firs died out .
Curve ( H ) shows the changes of thermoelectric force in a piece of bhe same kind of wire when in its original state , and ( K ) the modification which ulted from heating the wire to redness in a Bunsen and cooling it in air .
In both cases the effect of annealing is to depress the curve .
* ' Roy .
Soc. Proc vol. 65 , p. 230 , fig. 1 , 1894 .
moelectric Power produced by Magnetisation .
A glance at any of the curves of thermoelectric force for iron show how easily errors might arise if it were asstlmed that a wire which had been subjected to a magnetic ficld ecame perfcctly demagnetised when the field was withdrawn .
In all my the curves would have been much lower but for the demagnetisation by reversals before every observation .
The ratio of the wires which I used was generally so small , and the self-demagnetising consequently so great , that an apparent of force in strong fields did not often occur .
But in the case of a wire cm .
in length and cm .
in diameter , this reversal ttppeared in fields above 500 .
When the ( wire was ) jected to field exceeding this strength , there occurl.ed a galvanometer doflection to the right , indicating a genuine increase of thermoelectric force ; but when the magnetising current was rupted , the spot of light , ' of going back , went still further to the riglJt , the thermoelectric force due to the magnetism being greater that to the strong field .
the magnetising circuit again closed before tho wire was demagnetised , the spot of light , of lrse , moved to the left , and if the residual magnetism were disregarded , it wonld ) supposed that a reversal of thermoelectric .
The spurious reversal was also very conspicuous in the case of the steel wires . .
LXXIII .
2 Ir Dr. S. Bidwell .
On the Changes of \mdash ; Since the effect of magnetisation upon a nickel rod or is to shorten it , while , if the " " correction\ldquo ; be admissible , iron is always lengthened by magnetisation , it was to be expected from analogy that the effects upon thermoelectric power would be opposite in iron ) in nickel .
In iron the thermoelectric power of the magnetised respect to the unmagnetised mstal is positive ; in nickel , therefore , should be negative .
This view , though in accordance with what generally accepted the fact , is , however , at variance with the of all my experiments .
Five different samples of nickel were used , curves for four of these L ) , , ( N ) , ( O ) , eing given in fig. 7 .
Curve ( L ) relates to a piece of thick wire mm. in diameter , bought of Messrs. Johnson and Matthey as pure .
The retraction curvefor another piece of the same wirs is that marked ( h ) .
Curve ( M ) was giv by a wire mm. in diameter ; this was also supplied by Messrs. Johnson and Matthey , and described as being " " as pure as conveniently possible Curve ( N ) shows the result of an experiment with a sample of wire snpplied Iessrs .
J. J. Griffin and Sons , which did not purport to be and probably contains a considerable proportion of iron . .
Trans. ; vol. 179 , p. 228 , fig. 6 , 1888 .
Thermoelectric Power producecf by Magnetisation .
431 Curvs ( O ) was obtained from a strip about mm. wide and mm. thick , which was cut from a rolled sheet purchased at a metal warehouse .
Curve shows the retractio ] ] of a strip of the same meta19 mm. in width .
* The fifth specimen was a wirc from a piece of nickel gauze .
With this , too , the ised to he always thermoelectrically positive to the unmagnetised metal .
[ A sixth specimen , consisting of a mm. in diameter , has been recently tested , with same result .
, 1904 .
] The form of curve ( N ) which rises to a maximum at nbout suggested a possible source of error by which Thomson may have been misled , and I therefore repeated his experiment .
The arrangement which is shown in fig. , is essentially the same that employed by Thomson .
A piece of riffin 's impure wirc lvas bent into the shape of a horse-shoe , as shown , one of the limbs ough the small magnetising coil , the ellds of horse-shoe were ected by brass binding-screws to wircs to th galvanometer T. Heat was applied at or near by touching the wire with a hot copper rod , and when the magnetising circuit was closed , the galvanometer indicated a ourrent which llsually the direotion as in the other experiments\mdash ; from grletised to magnetised through hot .
It was , however , found to adjust the of the field and the position of so that a thermo-current flowed in the opposite direction .
This was the case , for mple , when the field-strength and the distance of from the coil 8 cm .
If a slight change were made in the position of , the ) lication of the field produced a current in the direction .
This deceptive effect , which , of course , really occurred betwcen more strongly and less strongly magnetised portions of the metal , could not be with the pure nickel wire , nor with the wire taken from the , which were the only other specimens tested .
With iron wirc , howcver , it quite easily produced .
Curve , fig. 7 , shows in ten-millionths ot the mcchanical compression due to ; in compttrisoll with that for iron , : oc .
cit. , p. , fig. 4 . .
S. Bidwell .
On the Changes of curve ( d ) , fig. 4 , it is seen to be quite insignificant : Since no experiments have been made to determine the lifting-power of nickel , this was calculated from the expression .
Values of I were obtained from a table given by Ewing , based upon results of an experiment by Du Bois ; the specimen used was an ovoid instead of cylindrical wire , but the values might be increased or diminished by 10 per cent. without influencing the corrected curve to any appreciable extent .
Young 's modulus was take1l as , approximating , the valus found for nickel by H. Tomlinson ; this also not be very accurately known for the present purpose .
Table IV shows how the ordinates of the compression curve , and those of the corrected curve , were determined .
Curve ( n ) which nearly coincides with the thermoslectric curve ( L ) is the uncorrected retractio11 curve ( k)inverted andplotted to the scale of ordinates in the right-hand margin .
If the corrected curve were plotted to the slightly different scale just within the margin , it lvould be substantially identical with , which may therefore be regarded as representing either the corrected or the mcorrected curve of change of length .
The small difference noticeable in the forms of the two curves ( O ) and ( l ) for the two nickel strips is just such as would be caused by the difference between their dimensional ratios .
As to the curves in fig. 9 , which show a remarkable qualitative correspondence with regard to the complex effects of tensile stress , nothing need be added to what has already been said .
The dotted curves are of course inverted .
* 'Magnetic Induction , ' edit .
, p. 161 .
'Roy .
Soc. Proc vol. 37 , p. 390 , .
Soc. Proc vol. 47 , p. 475 , fig. 3 , 1890 .
434 Power and Magnetisation .
[ Apr. 11 , and evidently cannot be made to resembls it by applying the correction for mechanical compression .
The compression curve ( r ) was constructed , like that for nickel , from values of I furnished by Ewing ; Young 's modulus was taken as , the value found by Tomlinsonfor unanmealed cobalt being The thermoelectric curve ( S ) was given by a strip of rolled cobalt , for which I am indebted to the kindness of Messrs. Henry Wiggin and Co. , of Birmlngham .
It is of interest as indicating ths similar behaviour of a very different sample of the metal .
If there is any relation between the thermoelectric and the strain phenomena ' in cobalt , it is obviously disguised by some cause which has yet to be discovered .
[ Note , 1904.\mdash ; Some further experiments have been made with specimens of cobalt which had been very thoroughly annealed .
For cast cobalt the change-of-length curve is entirely altered by annealing , becoming , at least in fields up to 1360 units , straight line represented by is the retraction in ten-millionths .
The thermoelectric curve is considerably lowered , but in other respects is not much affected .
No tion between the thermoelectric and the strain phenomena could be traced .
From an examination of the curves in fig. 10 it appears that the thermoelectric power of the unmagnetised with respect to the magnetised cast cobalt is proportional to the compressive stress , and consequently to the square of the magnetic induction .
] * ' Roy .
Soc. Proc vol. 39 , p. , 1885 .

