]\gt ; 351 XIV .
the Specific Resistance of Mercury .
R. T. GLAZEBROOK , M.A. , F.R.S. , of Trinity College , and T. C. FITZPATRICK , B.A. , Fellow of 's College , the vendish L Cambridge .
Received June 19 , \mdash ; Read Jnne , 21 , 1888 .
[ PLATE 19 .
] OF late years several determinations of the resistance of mercury have been made , and the differences between the results arrived at have been greater than would be expected at first sight from the nature of the observations involved .
The results of the experiments have been expressed either in terms of the ohm absolute C.G.S. units ) or of the B.A. unit , which , according to the determinations of Lord RAyLEIGH and one of the authors of this paper ( R. T. is equal to 98667 ohm .
In the case of Lord RAyLEIGH'S observations , a direct comparison was made between the mercury unit and the original B.A. standards .
Other observers have constructed copies of their mercury resistances in German-silver wire , which have been compared with the B.A. standards at the Cavendish Laboratory by one of us , or have compared their tubes directly with copies in platinum-silver wiro of the B.A. units which have been sent from Cambridge after careful testing .
The result of these various comparisons of recen years is as follows , and may conveniently be put in tabular form , giving the value in B.A. units of the resistance of a column of mercury 1 metre long , 1 square millimetre in cross section , at Centigrade .
This is done in Table I. , which also giyes the value as found by various observers of resistance of 1 ohm expressed in centimetres of mercury at C. 28 .
11.88 ON THE 5 3 should have by exactly the same amount during the last twenty-one years , it is inferred that no change has taken place in them , and the B.A. unit is defined as the of the values of the six coils at the at which they were said by HOCKIN to be was used in sense by RAYLEIGH in his electrical papers ( ' Phil. , \amp ; c and this is the attached to it in the various reports of the Electrical Standards Committee tl The method employed the tions d but little that Lord RAYLEIGH'S paper .
The resistance at Cent. of tlJe column of mercury tube is determined , .
units ; , of the column of mercury is ; its mean cross section is found by measuring a known temperature the column nearly filling the tube , and then the mass of mercury in the column .
The mean cross section thus found needs COl.rection for ) ularities in the tube , and these are obtained by the ordinary process of .
The .
as given in MAXWELL'S Electricity and , requires a nall Crection , for the fact that the length of the column used to ( letel-mine the cross section does quite fill the tube .
Let be the cross section at a distance from one .
Let be the ' of thread cury , which is passed the tube , when its middle point is at a distance from one end .
Then , assuming the cross section to be constant over the , we have , where is constant volume occupied by the read ; hence , if be the ) of points at which is measured the density of lnercury in grammes per , let be the average cross section of the tube over the tion ( at the end ch is not occupied by ] mercury used to find the average cross section , and let be the average cross section of the rest of the tube .
Then , ; . . . . . . . .
( 1 ) and , if ' be the resistan ce of a column of reury 1 1netre long , 1 square mm. in at Cenb .
, the measured resistance of the tube , ; there fore , 11DCC so we ( 4 ) ( 5 ) The term depending o11 is , of course , extremely small , but in some of the employed it exercised sensible effect on the result .
In addition to the corrections necessary to reduce the results to the standard C. , the lengths and rcq ctions of importance .
The extremities of the tube opened into two large cups which were filled with mercury , and ] observed resistance includes that of the mercury in these cups which is situated just beyond the ends of the tube .
Lord YLEIGH has shown that , on the assumption that the diameter of the mercury column in the is compared with that of the tubes , the correction required would be equivalent to adding to the length of the tube of the diameter .
The experiments of MASCABT , RVILLE , and BENOTT , ' des iences sir la De'termination dc l'Ohm , ' Paris , -Villars , 1884 , havejustified this theoretical conclusion .
To make more certain on this point , we decided to repeat MASCART'S experiments .
A third was , of the same size as the two terminals which described later , from them only in having two , such as , Plate 19 , fig. 2 .
A tube , cm .
long , and .
bout 1 .
unit in resistance , was taken .
Its resistance when filled with mercury in the usual way was found .
Its diameter , as found from its length and , was mm. ; and therefore the theoretical correction for the two ends is equivalent to an addition to the length mm. , or to the resistance of .
unit .
The tube was then cut in two pieces , and an end was inserted in each of the two openings in the third mercury cup .
This was filled with mercury , orked up , a neter passing the cork , and replaced in the trough with the usual packing of ice .
When the whole had cooled down its resistance was again measured and found to have increased by .
unit .
This increase is due to the tance of the two cut ends , and the difference in the observed and the theoretical is within the errors of the determination .
Another determination was made with a tube of about twice the cross section .
In this case the correction was equivalent to increase of resistance of .
unit , while the observed as 000 .
It will be noticed that in both cases the value was less than the theoretical .
This result was also found to be the case by Lo , who gives as the factor in the correction deduced from his expel.iments , the value being diameter of the bore , and the SPECIFTC outer diameter of the ) .
In the case of the last men tioned mm. , and the ) , so to , for ; tube , the correction be .
Our ; wonld this to be too small , though there is some evidence for a rather than 82 .
At the same time , it is to justify any change , .
nd we therefore add to the observed of the tube a quantity SL equal to 82 of the diameter of the tnbe .
The correction to the length is not quite so simple .
It arises from the ends of the column are not plane at angles to its length , but ; of a curved surface which is spherical only if the tube , which was placed in a horizontal position , be so narrow that the effect of crravity may be eglected .
length is the extreme length of the mercury column meas lred f end to end , and the volume found from it is therefore great by the amount contained between ) mercury and two vertical planes touching the mercury column extremi tie of each meniscus respectively .
This may be expressed in fornl .
Sl , where is the radius of the } and a correction to be subtracted from the length .
In ) in which the end of the mercury is spherical the calculation of S1 is simple .
Let repl.esent a section of the mercury meniscus by a vertical through the axis of the tube ; let CD be the axis of the tubc , meeting at angles in ; let the angles of contact at A and , between the and the glass of the tube , be ; and let be the centre and the radius of the mercury bubble .
Then and Angle DAO The length of the mercury column was determined by read i microsccpes , be lained below , and in all cases the readings for both A and were taken so .
R. T. BROOK AND T. C. FITZPATRICK the distance ) was measured .
This should , of course , be cons ant for any given tube , and ) iment s it was nearly so .
Let us put Then we AD OA ITence , .
Now , is known with sufficient accuracy from ) and ] of the cury , so that the above equation gives us The Table gives values observed : TA mean , allowing for .
number of observations in each tube , is .
The Igle of contact between mercury and glass is usualJy given as , so the reement is very good .
Now , volume of the spherical segment ACB is easily seen to be , that of the cylinder on AB as base and of height CD is .
The correction : therefore , to be subtracted for each of the tube from the whole lume is the difference between these two , and this ; erefore , ON SPECIFIC OF MERCURY .
o ) If we pnt , as found above , then and this is the valne used for the tubes VI .
, VIII .
, and II .
in the Tables which follow .
In the case of the wider tnbes , it was not at first so clear what the amount of this correction ought to be .
If ACB , fig. 2 , again esent the meniscus , it evident that was not vertically below further , the extreme point did not lie on the axis of the tube .
Let AE be vertical through , and a verCica ] the end of the meniscus , and let be the axis of the tube .
Then , in the case of one tube of mm. in diameter , we found AF mm mm , , approximately .
) ations on other tubes gave somewhat similar values , { he di erence between BG and AF was not always so great as in the above .
The exact calculation of the between the plane through FG and the eniscus ACB is not possible .
We have calculated the correction on assumption it is the same as that for a spherical through AC'E ; the effect of gravity has been to draw this surface down into the position , and it is assumed that the volume is tely the same as AC'E .
Taking the values given aboye , we find on this It seemed desirable , however , to result by direct expel.illlent , ] did , following the metl ) adopted by Lord .
Ebonite plugs were turned which exactly fitted the ends of the tubes , and these .
were inserted and cssed up against the ends of the mercury columns so as to flatten them , the of the column was measured with the reading microscopes .
The ebonite plugs then , and the length of the mercury column measured .
The erence between these two gives us Sl directly ; for one tube mm. in diameter mean of a number of determinations which were in fair agreement gave mm. ; and for this ) we therefore , For a tube snch ( those used for the half units , for which the tmeter was mm. , we found mm. , an this gives It is , therefore , clear ttlab for these tubes.we may , without serious error , use the value given by the above theory , has been done in the -lions .
Thus , the equation to determine becomes ; ( 6 ) thd , will be seen when the of the various quantities involved are introduced , this may be writterl . . . . .
( 7 ) In this expression temperature corrections necessary to , while the weight will require reducing to its value in vacuo .
Let temperature at which the length is measured ; coefficient of linear expansion of meaSUl.ing rod ; temperature at which the thread of length is measured ; density of mercury .
grammes per .
; coefficient of expansion of mercury ] ; coefficient of cubical expansion of glass *These lues a taken from the ' Travaux et Memoires du eau International des et Mesures , ' vol. 2 .
See ' Nature ' for April.3 , 1884 .
ON THL OF RCURY Hence , Volume of thread at \mdash ; \mdash ; lie lSIean section of at Mean section of tube at \mdash ; \mdash ; Length of tube Hence , the value of corrected for rature is ; and , if be the weight density of air , of bhe br ss used , taking dry air at , the mean tenlperatule of the weighing and ] .
Hence , finally we get , the cections , \mdash ; G2 \mdash ; ; ( 9 ) and , if we put in nerical values for tho one unit tubes , we have 0149 ; while , for the other tubes , the third lerm is .
We may , ( 10 ) where is a small fraction , being the sum of all the .
terms with their proper The methods used in but little from those ernployed by Lord .
The tubes were supplied by Messrs. POWELL and .
) and a number were ) uf \#8218 ; \micro ; selected and were cut so as .
R. T. BROOK AND T. C. FITZPATRICK to have nces of vely approximately ] .A .
unit , 1 legal ohm , .
unit , and .
unit resl ) ectively .
ends of the tubes were )rotlnd square by placing each in groove in a piece of hard wood and allowing the end to pass through a hole which it ] fitted in britss } ) late fixed at right angles to the groove .
The tubc to rotate slowly round its axis , and the end ground by emery on a copper ] ) find the value of the length ( L ) , two smaall rectangular pieces of brass were used .
These were ] squared , alId 1 fine xved on a small piece of white inserted in the centre of one of the faces of each block .
The tube was placed under two reading microscopes , which could be adjusted by micrometer screws , graduated to , and of the rectangular pieces of brass was brought np to each end , and adjusted so that one edge touched the tube as nearly as could be along the horizontal diameter while the cross mark on the lay on the axis of the tube produced .
microscopes we re then focussed on the crosses , and several dings taken as the tube was turned round on its axis ; except in the case of one tube ( No. III .
) , no appreciable difference was observed on the tubes round , and in the case of III .
the difference did not amount to timetre .
The tube and brass pieces were then removed their taken by the standard metre of the Cavendish This is a bronze rod 1 metre long , divided into decimetres ; the first etre is divided into .
The len of the rod , as determined at the Standards Office of the of rade by comparison with the standard metre , S.S. , is cm .
at C. , while its coefficient of expansion is 000017 per 1o C. The first decimetre and the millimetre subdivisions have no appreciable error .
For the lengths greater than one metre a second bar , made by the Instrument Company , wris used .
This cm .
in ength , and was divided into decimetres , which were compared with those of the standard .
The two bars were placed end to end in a long wooden box , which they fitted htly , and the distance between the division of the standard and the first ision of the bar measured with the microscopes .
The temperatures were read by a fhermometer laid on the scale or against the glass tube .
The length thus found was that of the glass tube , together with the sum of the distances between the cross marks on the brass blocks and their edges which were in contact with the tube .
This distance was carefully determined by aid of the reading microscopes and the standard metre .
values found for this correction differed at most by 005 cm .
, and the observations on the of the tubes , which were repeated in each case three or four times on separate occasions , agreed to about the same amount .
Thus , the error in the length of the tubes is probably in no case greater than cm .
in a of about 100 cm .
ON THE SPECIFIC RESISTANCE OF MERCURY .
measured in various positions close up to the end .
The same process was repeated at the other end : let the mean of the lengths thus found be .
Then , if we suppose that the curvature of the meniscus did not alter ( and experiment showed that this assumption is nearly true ) , we have very approximately the equality Thus , in tube VI .
, which is slightly smaller both ends than it is in its central part , the value of is mm. , while the length of the thread was the same at each end , and was equal to mm. For this tube the mean value of was 00367 , and the average value of the correction .
It is larger in the case of this tube than with any other used , in most the effect of one eIJd was opposite to that of the other .
Table V. gives the values of ] engths of the threads at the two ends respectively , the mean of these two or , and for the various tubes .
TABLE V. end .
The mercury was weighed , using a balance by OERTLING and the weights employed by Lord RAYLEIGH ; these were compared with each other and with a set of weights which Mr. SHAW had previously compared with the standard 500-gramme weight of the Laboratory .
A small correction of about 1 milligramme on 10 grammes was found , and has been introduced , but it is too small to affect our results .
The mercury was weighed in both pans of the balance , and the weighings repeated on two or more different days .
In the electrical measurements the tubes were compared directly by CAREY FOSTER'S method with the B.A. standards , using the bridge designed by Dr. FLEMING which was employed by Lord RAYLEIGH .
The tubes had been so adjusted that the difference between them amounted only to a few centimetres , at most 70 divisions of the bridge wire .
We were thus independent of variations in the resistance of the wire due to temperature changes , and the value of the bridge wire division was taken as .
unit .
The ends of the mercury tube were connected , in a manner to be described shortly , SSRS .
R. T. GLAZEBROOK AND T. C. FITZPATRICK the bridge by copper rods , and rods of the same material and almost the same resistance were used to connect the st , andard coils .
It was hoped in this manner to compensate the effect produced on the resistance of these rods by changes of temperature in the room .
Since the rence between the two sets of rods was only equivalent to one bridge wire division , this was completely secured .
The coils used were the following : the tubes VI .
and I. the standard was employed ; for V. and IX .
the standards and in multiple arc ; and for I. and III .
, and Flat in multiple arc .
In the case of the legal ohm , bube II .
, a coil of 100 B. A. units , ELLIOTT No. 68 , was placed in multiple arc with ) tube , and the difference between the combination and found in the usual way .
The temperature of the water baths in which the coils were placed was taken with a thermometer which had been compared at Kew , and the necessary corrections applied .
The temperature of the baths and of the room in which the experiments were made never differed greatly from The following able V. gives the values of the coils at , with their temperature coefficients .
value of and are taken from Dr. FLEMIN hart .
The difference between the two coils at the time of bhe observations was determined and found to agree exactly with that given by the chart .
In the case of Flat , which is not one of the six coils mentioned on p. 352 , repeated observation during the last two years has shown that it is now slightly lower relatively to the others than when examined by Dr. the change is not greater than0001 B. A. unit , aod is probably due a slight imperfection in the insulation .
We have taken the value relative to and given by our own observations ; as Flat is only used in multiple arc with and for the tubes I. and III .
, an uncertainty in its value is divided by nine in the result , and the error introduced is too small to trouble us .
The coil of 100 B.A. uIJits is one of the of the Association , and , like the cther coils , is of platinum-silver wire .
TABLE VI Coil .
FIat No. 68 For the one unit tubes the terminals of dipped into mercury cups on ebonite , which were connected to the bridge by the copper rods above mentioned .
When the coil was used its terminals dipped into the same cups , and for the one-third unib tubes I. and III .
, the coil Flat dipped into two other cups connected with the first two by thick pieces of copper .
The resistance of these connexions was determined ON THE SPECIFIC RESISTANCE OF MERCURY .
by iinding the difference between Flat and directly , and then when Flat was connected to the bridge by these copper pieces .
In this way we found the resistance of the connexions to be .
unit .
The temperature was about the same as that at which the connexions were used .
In the values of Flat given in Table VIII .
the resistance of these connexions been included .
In all cases the temperature of the room was almost exactly the same as that of the water baths .
One of the ebonite cups into which the ends of the tubes opened is shown in figs. 1 and 2 , Plate 19 , which is drawn to scale full size .
In their design two points mainly were attended to .
The first was , that it should be possible to reduce the mercury them very nearly to C. ; the second , that there should be no contact between copper and mercury , for BENOIT has shown that the conductivity of mercury is in a very short time appreciably increased by contact with copper .
The glass tube passes through an india-rubber cork , which fits into the terminal at ; the tube was usually adjusted so that its end was flush with the inner surface of the terminal .
Mercury was then poured into the cup and allowed to run slowly through the tube into the second terminal until each was about two-thirds full .
The top shown in fig. 1 was then placed over the terminal and secured by four small screw-bolts passing through the flange .
When these were screwed down the terminals were completely water-tight , and could be left covered with melting ice or water fur days without leakage .
The top consists of a flat plate of ebonite , with four holes to receive the bolts .
Through this plate two ebonite tubes , ( figs. 1 and 3 ) ) .
A hollow platinum cup , , about 35 cm .
long by ratber more than 1 cm .
in diameter , is secured firmly into the tube , and a thick piece of copper rod , , fits the .
of the cup tightly , any interstices between the two being filled with mercury ; surface of the copper rod was well amalgamated .
This copper rod , , is brazed to the coppel rods , , which form the connexion with the bridge .
Pieces of india-rubber tubing , , about 10 cm .
long , are fastened over the upper ends of the tubes , ; the connexion to the bridge passes through a cork , which closes upper end of the tube , the junction is made water-tight with marine .
A thermometer , , graduated to fifths of a degree Centigrade , passes through and the india-rubber tube , , which fits it closely , alud gives the temperature of the mercury in the terminal .
In taking the observations the ice was packed closely round the terminals up to the tops of the tubes , and , so that the copper rods were su1Tounded by ice for 12 or 14 cm .
above the level of the mercury .
Contact between the copper and the mercury in the terminals vras thus established through the platinum cup , ; the surface of this is about 12 sq .
cm .
This surface was amalgamated in the followmg manner : cups were platinised by electrolysis from a solution of platinic 0066 SSRS .
R. T. GLAZh'BROOK ANDJ T. C. FITZPATRICK chloride in nitric acid .
On immersing the cups in mercury , after heating to drive off traces of the acid , nalgamation readily took place .
After this process thoroughly good contact between the platinum and the mercury was secured .
To test this , the two platinum cups were placed in the same vessel of mercury , and the ends of the copper rods connected with the bridge .
In this position the resistance of the connexion was measured , and gave the same value before and after the experiments , while no appreciable change could be noticed on taking one of the platinum cups out of the mercury and again replacing it .
When only nbout onethird of each platinum cup was in the mercury , the resistance was increased by about .
unit .
In use care was taken to place sufficient mercury in the terminals to cover the cups entirely .
The following additional experiment shows the goodness of the contact : The platinum cups were aced in the mercury , and the resistance measured as described ; then the copper rods , , were removed from the interior of the cups and placed in the same veesel of mercury , the other ends of the copper rods being in connexion with the bridge , and the resistance was again measured .
No difference between these two measurements could be detected .
Thus , the contact through the platinum was practically as good as if the copper rods , , had dipped directly into the mercury in the termina]s .
Tables VII .
and VIII .
give details as to the mercury nployed .
The genexal method of treatment was as follows :Mercury from the ordina1y stock in the Laboratory was treated with nitric acid and potash , and then distilled in vacuo in the Laboratory still .
his mercury after bein once distilled was again mixed with nitric acid , being allowed to stand overnight in colltact with it .
After this it was heated with caustic potash , and then well washed and dried by being strongly heated , and finally it was passed through a second still , newly set up for the purpose , in which only mercury which had been previously distilled and treated as above was ever placed .
Tables VII .
and VIII .
, which contain the results of the observations , do not require much explanation .
The one unit tubes are given in Table tube VI .
ffiling 2 , the curvature of the ends of the mercury column was much less than in any other case , the correction has , therefore , been calculated specia ] for tube The result is clearly too high , but there is no reason in the details of the measureents for omitting it .
It may be noticed that the results of the 4th and 7th fillings are not given in the Table .
These will be referred to again shortly ; they were fillings of a special charactel , and it seemed best to treat them separately .
The extreme difference between any two fillings is 000135 , and the difference between the mean result and the extreme is about half of this .
In some of the columns the figures have been given to six places ; the only object in this is to secure accuracy in the fifth figure in the final result .
In the Table for tube VIII .
, the fourth and sixth fillings are omitted for a similar reason to that given above .
The extreme difference between any two fillings is about the same as in the case of VI .
The mercury used in VIII .
5 was taken directly the iron bottle in which it was supplied by Messrs. TUBBS and WILKINS , and was not distilled in the Laboratory .
In the first filling of VIII .
, and the first and third fillings of , the filling from which the cross section is determined was different from that used to determine the resistance .
This was due to the that at first the necessity of having the tube almost completely full when finding was not fully appreciated , and too much mercury was allowed to escape from the ends in removing the terminals .
In the other cases in this table the two fillings were the same .
The mean value for tube II .
is reduced by the result of the 2nd ] , which is clearly too Iow , but the observations themselves do not give us any reason for rejecting it .
All the fillings for this tube have been included .
The mean value of found from these three tubes is .
unit , and the greatest difference between any two observations is 000241 B.A. unit .
The average error independent of sign is 00005 .
If , however , we rejected the results of VI .
2 and I. 2 , the mean would hardly be ffected , but the mecm error would be reduced to 000037 .
We may , therefore , fairly put as the result of the observations on the one unit tubes .
This is , it will be observed , bhe value given by tube VIII .
The value for .
is raised unduly by the observation VI .
2 ; that for II .
is unduly lowered by the result of II .
2 .
R. T. GLAZLBIIOOK AND .
C. FITZPATRICK on the and one-third unit .
of T. , which is not given , was taken , but it was clear that been made in the urement of of for the result the others by over one in a .
This is the only ] in the has ) entirely ] ejected .
The result of V. 1 is too If result of the ] in VIII .
equal weight , we obtain vadue B. A. unit , and the ayerage error is 000029 .
Gllce ) even thd the resnlt derived the one unit tubes is , at first , considerable , but it must be remellbered that for the tubes I. and III .
an or in measurement of resistance of 00003 or one thi -thousandth of a B.A. unit risc to ] error of one ) ndth i the result .
This , of course , is an elncly small quantity .
Morcover , the small uncertainty which attaches to the rrections SL Sl would a larger effect in these large tubes , and our observations tend to show that the coefficient 82 in the yalue for is possibly too Jt will be lloticed , howevel , that the values of actual observed resistance of tube , .
among thelnselves by extremely small quantities .
In taking a final , it was clearly unfair to weight the observations equally , and to the conclusion the probable accuracy was roughly inversely proporI the ) of cross section .
We have , , attached the to the esults from the one unit , half unit , and unit tubes respectively , at the nlt esistance of a of etre long , 1 square millime{re in section , at is .
unit .
lf we equad eights to tiollS , the result will be 95351 , so the effect of is hardly ppreciable .
It to consider the four special fillings of tubes VI .
and VIII .
, which been omitted .
In two of these , and VIII .
4 , the attempt was made to fill the quite full , when ; a very small bubble of mel.cury was left protruding from each end of the tube when it placed under the reading microscopes , and flat of brass were ) simultaneously against the ends , there being .
thin ) betweell the brass and the mercury .
It was hoped in this to out ] erfluous mercury leave a column with ] flat ends exactly filling .
It eems p ] in the case of VI .
4 that this successfully ccon)plished , the v of found from the experiment is With , however , it cleal , on looking through the the ( in contact ) ) vere slightly curved , this still Jnore obvious when the brass papel were emoved .
The resulting of is ccol d too being .
unit .
of the two and VIII .
6 , mercury was used which had ON THF , RESISTANC OF CUIlY .
been passed once through the still at the University Chennioal boratory , and then treated with nitric acid .
It was clear , from the ) of the , that it was impure ; but it was thought of interest to determine a vadue .
mercury in purifying of which no special trouble had been taken .
The rity shows itself at once in the results , for , while the mean value of for VI .
is , for this ] ; while for VIII .
the resistance of the fillmg is .
unit , against a mean of .
The corresponding values for ?
are This mercury was then treated with nitric acid , \amp ; c. , redistilled in our own still , nd used again in the fillings VI .
8 and VIII .
7 ; the values of Fu were ) ] , and of and respectively .
Thus , the impurity has been clearly removed by the distillation and acid treatment .
In some cases a tube was filled on one afternoon , and its resistance determined .
The whole was then allowed to stand over , being re-packed in ice in the morning , and the resistance again measured , but no appreciable lloted .
Thus , for VI .
, on January 9 , the value 99994 was found , while the same , re-packed on January 10 , after the mercury had stood for 16 or 18 hours in contact with inum of the , gave ; the difference is within the temperature errors of the coils .
Some experiments were made on the of known impurities on the mel.cury in altering its resistance .
In one case , about one two-thousandth part of zinc filings was added to the mercury .
On mixing , the surface of the mercury was made foul ; the mercury was then passed through a filter paper and used in VI .
, but the effect on the resistance was not appreciable .
It is probable , of course , that the filtering had removed a large portion of the zinc , but the experiment gives some idea of the amount of lrity which the resistance measures will detect .
In another filling of VI .
, a mixture of , with a small percentage of tin used .
The resistance was much too small to be measured on the bridge\mdash ; the bridge wire has a resisbance of .
unit ) .
This mercury was treated in the usual way , distilled in the Laboratory still , and then in our own , and on being again used gave as the value for B.A. unit .
fhus our treatment was sufficient to remove the tin from the mercury .
Some obseryations were also made on the change of resistance of lIlercn ) temperature .
Thus , on January 5 , tube II .
was placed in a trough in water at about the ture of the room , and its resistance measured .
The tube was theu packed in ice and measured .
Similar observations were made with tube III .
, and the results are givell in the Table IX .
Other observations confirmed the results there given .
ON SPECIFIC ANCE OF MERCURY .
Thus , STRECKER'S value is higher than the mean by 00003 , and if we were to reduce his observations from to , using the mean coefficient given above , we should obtain the value for , 95362 .
This value may possibly be a little too high , but at any rate the reasoning is sufficient to show that the difference may easily depend on a small error in the ture coefficient .
The same reasoning will apply to KOHLRAUSCH'S results , for his comparisons between the mercury tubes and the wire standards were usually made at temperatures differing from , and were reduced to by the use of STRECKER'S formula .
The value of the ohm in centimetres of mercury at , as given by KOHLRAUSCH in a letter to R. T. G. , February 16 , 1888 , is , agreeing exactly with ROWLAND .
MASCART , NERVILLE , and BENOIT found a value for which is as much above our value as KOHLRAUSCH is below it .
At the same time , their value of the B.A. umit in ohms is less than ours , leading to the result that the value of the ohm in centimetres of mercury is ; or , again , the same value as ROWLAND'S .
This might seel1l to show that there was some small change in the B.A. unit used by MASCABT between the time it was compared at Cambridge and bhe date of their observations .
[ Qnite recently , June 4 , 1888 , M. WUILLEUMIER communicated to the Academy of Sciences at Paris the results of some experiments by LIPPMANN'S method , which give the value ] Thus we may conclude that the experiments of MASCART , STRECKER , ROWLAND , KOHLRAUSCH , WUILLEUMIER , and ourselves are in fairly close agreement , and thnt * WrEDENANN , Annalen , ' vol. 25 , p. 475 .
'Electrotechn .
Zeitschr vol. 3 , 1882 , p. 408 .
'WIEDEMANN , Annalen , ' vol. 25 , p. 11 .
S 'Etudes Electrometrologiques , ' vol. 2 , 1884 .
" " Resume d'experiences sir la determination de l'ohm , \ldquo ; ' Journal de Physique , ' 1884 , p. 230 .
'Phil .
Trans 1883 , p. 185 .
' Phil. Mag October , 1885 , p. .362 .
, p. 372 .
The results of Table are represented graphically in fig. 3 , in which the abscissae represent temperature and the ordinates resistance , the temperature being that indicated by the thermometers in the cups .
The tube was of course packed in the ice during these observations .
It would appear from the curve that our own observations may be slightly too high , possibly as much as 00004 , through the temperature in the cups being on the average C. instead of , while at an error of about might be introduced .
This error is equivalent to that caused by the whole tube being at instead of at , or by about 5 per cent. of the tube being at the temperature of the mercury in the terminals .
In tube VI .
some 6 or 7 per cent. of the tube was within the corks used to close the terminals .
It may be noticed that the observations on tube VI .
given in the last line of Table XI .
were made with Lord RAYLEIGH'S terminals .
We thus infer that , while the fact that in Lord RAYLEIGH'S experiments the terminals were at or may explain a small part of the difference between our results , reducing his by about 00024 , it cannot possibly account for the whole , mounting as it does to 00060 , and we must look in some other direction for the explanation .

