V. " An Account of Experiments made at the Kew Observatory for determining the true Vacuumand Temperature-Corrections to Pendulum Observations .
" By BALFOrUP STEWART , Esq. , F.R.S. , and BENJAMIN LOEWY , Esq. , F.R.A.S. Received May 27,1869 .
1 .
Pendulum-observations , whether undertaken for the purpose of obtaining unalterable standards of length or for physical and geodetic objects , are usually made in air or in a receiver , from which the air is partially or almost entirely withdrawn ; and in order to render such observations , made at different places and by different observers , capable of intercomparison , they are , by means of a " correction for buoyancy , " reduced to a vacuum .
It is well known that the most illustrious physicists and mathematicians have given a great deal of attention to a correct determination of the principles on which this reduction to a vacuum ought to be based , and of the actual resistance which such a body as a pendulum meets during its vibrations in a fluid body .
Until some years ago , especially since the researches of General Sabine and Bessel , it was thought best to determine for every pendulum a certain constant by finding its vibrations in air at the usual pressure , and also in a receiver from which the air is as much as possible withdrawn ; from the difference in the number of vibrations thus found the correction was then calculated on the assumption that this difference is proportional to the difference of density of the air .
2 .
In the pendulum-observations made at the Kew Observatory in connexion with the Great Trigonometrical Survey of India ( vide Proceedings of the Royal Society for 1865 , No. 78 ) we adopted , for determining the necessary constant , the method first carried out by General Sabine , and of which a detailed account is given in the Philosophical Transactions for 1829 , Part I. page 207 &amp;c. But since our account has been published , two eminent physicists , Professor Clerk Maxwell and Professor O. E. Meyer in Breslau , have independently investigated the internal friction in gases , and its effect upon bodies moving in them ; and among the prominent results obtained by them is this , that the influence of the internal friction of a fluid on a moving body is not proportional to its density .
However , for small differences of pressure , such as those experienced by General Sabine in his researches , the old method for determining the correction is sufficiently accurate ; or again , if a series of such experiments as our own fundamental Kew observations for India be made at a very low pressure , say from 1 an inch to 1 inch , the correction is itself a very small quantity ; and the application of a more correct principle of reduction will not sensibly affect the ultimate results , because the difference between the true and approximate correction is in such a case extremely small .
But if , as is the case in the Indian observations , experiments are made at higher and varying pressures , it is very desirable to apply experimental methods which will give the true correction .
3 .
With a view to collect for the theory of the subject a great many carefully conducted experiments , and also to supply those who are actually engaged in pendulum-experiments at the present time with practically valuable results , we proposed to ourselves to observe the behaviour of pendulums of the different forms hitherto used in such researches in which the pendulum is employed , at pressures varying through the whole range , from the lowest obtainable in a receiver to the usual atmospheric pressure .
The carrying out of our intentions met , however , with many delays through unavoidable circumstances , and there is , indeed , at present little prospect of our being able to complete the whole of the original plan .
We give , therefore , here an account of some preliminary results which are , in our opinion , not without practical importance , and which will certainly find their use in the reduction of observations made with pendulums of a form similar to that used by ourselves , viz. that form of reversible pendulum known as " Kater 's pendulum .
" 4 .
The following is an account of the operations:-The pendulum was swung in the Kew receiver , made of five pieces , two of metal and three of glass , the parts fitting closely , and the whole being connected with siphongauge and air-pump by tubes .
One of the metal pieces is perforated behind and in front , and the apertures are covered by plate glass for the observation of the coincidences .
The pendulum was swung at the following pressures : I. At about of an inch V. Between 4 and 5 inches .
' X. At about 20 inches .
( lowest obtainable ) .
VI .
, , 5 , , 6 , , XI .
, , , , 25 , II .
Between Iandz inches .
VII .
, , 7 , , 8 , , XII .
AtthefullatmospheIII .
, , z , , 3 , , VIII .
At about io inches .
ric pressure .
IV .
, , 34 IX .
, , , , 5 , , At each pressure a good many observations were made , in order to ensure reliable mean results .
5 .
With reference to the registration of the observations , we have strictly adhered to the method previously adopted after careful consideration , and explained in our former account ; hence we need not here enter upon this part again .
Instead of registering one coincidence at the beginning , during the progress , and at the end of an experiment , we have this time in most cases observed three successive coincidences , and the arithmetical mean of these , together with the mean of the corresponding registrations of arc , temperature , and pressure , stands for one observation ; we think that this method ensures greater correctness , although it is more laborious than that previously adopted .
6 .
The reduction of the observations comprises , as shown in the previous paper alluded to above , several corrections to be applied to the number of observed vibrations ; we shall mention here only those points which differ numerically or experimentally from the numbers or methods explained in that paper , which contains also an experiment with its full reductions .
By referring to these and the following remarks our method of procedure will be so abundantly clear , we hope , that we shall be able to proceed immediately afterwards to the statement of the final results .
2o 2 489 A. Correction of the observed arc-readings and reduction of the vibrations to infinitely small arcs .
We have previously shown that if D-=distance of scale from the object-glass of the telescope , d=distance of scale from the tailpiece of the pendulum , 0-=observed scale-reading for the whole arc of vibration , S=distance of indicating-point of the tailpiece from the knife-edge , a=true semiarc of vibration , O(D-d ) then tan a ( DS2DS expressing all distances in inches into which the scale is divided .
In our case repeated measurements gave the following mean values for these quantities : D= 10186 inches ; d-=056 inch ; S-47-55 inches .
D--d 1013 Hencewehaveto add log( =log(x 17 )=20194252 to the logarithm of the observed scale-readings to obtain the semiarc of vibration .
For the reduction to infinitely small arcs , we have again used the wellknown formula number of infinitely small vibrations M sin ( a +a ' ) sin ( a-a ' ) =n n. 32 ( log sin a-log sin a ' ) ' the symbols having the same meaning as previously stated .
We are well aware that more convenient formulae , and more correct methods , have been used or proposed by different observers for this correction ; but we thought it best to adhere to a uniform method in the reductions , in order to facilitate any future rediscussion of our original observations ( which are preserved at the Kew Observatory ) , should such appear desirable when the results of the Indian pendulum observations will be published .
B. The precise determination of the rate of the clock might have been of minor importance in our experiments , an approximate uniformity of rate being the chief desideratum .
We expected , however , that very small differences in the number of vibrations would result in those experiments where the pressure differed only by an inch or even less .
We considered it hence of the utmost value to have a precise record of the behaviour of the clock during these experiments , so as to discover at once changes in the rate , and to make our reductions depending on it for each experiment .
A great number of transit-observations were accordingly made , and during these not only the pendulum-clock , but also the behaviour of a chronometer by Dent and a meantime-clock by Shelton was accurately determined .
These two latter served during those days when no transits could be obtained for deducing the rate of the pendulum-clock by intercomparison .
From the whole of these observations the following Table of the number of vibrations during a mean solar day has been calculated for every day of 490 [ June 17 , the four months during which the experiments were carried on , each result being of course employed for the pendulum-experiments of the corresponding day .
The Table shows that our plan was the safest , as differences of nearly one second are observable ; these differences have , however , apparently no connexion with changes of temperature , as the rate of the clock during the artificial heating of the pendulum-room , of which we shall soon have to speak , showed hardly any difference from the mean rate , proving that the compensation was not faulty .
The pendulum-clock showing sidereal time , the Table is calculated from the formula:-Number of vibrations in a mean solar day =N'= 86636-5554 1--86-t ) ; and hence : Number of vibrations of detached VN ' pendulum =N= V , where V , V ' are respectively the number of vibrations of the detached and clock-pendulum from beginning to the end of our experiment .
TABLE I. Number of Vibrations made by the clock-pendulum during a mean solar day .
1866 .
September .
October .
November .
December .
1 .
76577'40 86577'38 86577'43 86577'50 2 .
" .
'25 '32 '40 3 .
s'5 *4 '20 '39 4 . ? ?
29 'o 86576'99 '48 5 .
86576'97 '21 86577'-0 '31 6 .
'83 22 8657694 '39 7 .
'8I '-6 '95 '50 8 .
'88 8657691 '99 '54 9 .
g90 86577'00 86577'i1 '54 xO .
86577'04 ' '1iI7 '51 1 .
0oo0 *3 '24 '6r I2 .
'07 '28 33 '68 13 .
'26 '*0 '28 '70 4 .
'i6 '04 '40 '70 5 .
' '22 53 '64 I6 .
08 '29 *6I '64 17 .
86576-98 40 6o -6o I8 .
'90 50 o 70 '70 19 .
70 '54 '75 '7I 20 .
'90 '6o '71 '66 21 .
'99 '6I '80 '59 22 .
'93 '43 '74 '4 23 .
'85 '28 70 '47 24..83 '27 62 '44 25 '99 '47 '5 ` 46 26 .
8657720 40 '49 '48 27 .
x '29 '44 '47 28 .
33 '33 64 '59 29 .
30 35 '56 6 30 .
8657724 '29 86577 56 '5I 31 ... .
86577'40 ... 86577'56 C. Correction for temperature , -The method of suspending the thermometers was precisely the same as that used in our previous experiments and described in the account we gave of them to the Royal Society .
The formula employed for deducing the most probable mean temperature for each experiment was as before , using the same symbols:(t+ t ' ) ( ?
' ?
+ t")i ?
( nj ' tl ) t to : +22n Jr n+ n ' * * cess of reducing our observations , came to the conclusion that it would be best to exclude all experiments made at temperatures above 100 ?
, and also those where great differences in the readings for temperature occurred , from our final results .
The principle which guided us was not to vitiate good observations by doubtful ones ; and the following small Table , showing the temperature-readings during four experiments , taken quite at random , will show best how we proceeded : I. Thermometer Th in front .
II .
ermometer behind .
Thermometer in front .
Thermometer behind .
Upper Lower Upper Lower therm. therm. therm. therm. 71-20 70.20 70'0 68'5 7I1I5 70-20 70'4 69'4 71x30 70'00 70'6 697 71'85 69-60 7I'4 71'4 72o00 69'55 7I'4 71'6 71'90 69'50 7I'4 70'7 Mean 7c'9 70'6 70'8 III .
99'5 98'7 92'4 9I'5 97'4 97'0 9I'6 90'6 95'5 94'3 90'3 88-I 82'5 79'9 78'5 76'2 ( 8I-6 79'3 77'4 760o 80-9 79'2 75'5 74'I J Mean 88-8 83'7 86-2 Upper Lower Upper Lower therm. therm. therm. therm. 85-6 83'7 84'8 83'5 86-i 83'8 84'0 83'0 86'7 83-6 83-3 82z6 78-4 78'8 82-2 8z-i 79'4 80'4 80'8 8I-z 8 -2 82'0 79'4 82-8 825 82'45 82'5 IV .
o8'4 107'3 ioo-6 99'5 o1076 io6'5 99'7 99 ' ?
0 105'7 103-8 98'5 97'3 97'4 96'5 93'I 9I'3 96-3 95'1 92'2 90'9 95*9 94'7 91'8 90'3 O11-3 95'3 98'3 Although in the rejected experiments the means of all readings of both sets of thermometers approach each other , still there occurs a fall of nearly 20 ?
at the end of an experiment , as compared with that temperature which is recorded at the beginning , besides differences of nearly 8 ?
between the thermometers in front and those at the back of the pendulum .
That the temperature of the latter during an experiment is represented by the arithmetical mean of such discordant readings , we think most unlikely ; and hence these experiments and similar ones were not used , although , of course , the number of available experiments was thereby reduced .
The following experiments , which represent the final results of this temperature-investigation , deserve at least some confidence , although we 4a 00 pq I0 Nl3 1869 .
] Corrections to Pendulum Observations .
493 should have liked to see their number much increased .
To avoid a correction for pressure , we took care to correct at once , before beginning an experiment , the reading of the gauge to 32 ?
of temperature , and to regulate by a few strokes of the pump the pressure , so as to assimilate it to the mean pressure ( also reduced to 32 ?
) of the experiments previously made in cold air .
All pressures are reduced to 32 ?
, and the small difference of pressure which still resulted , comparing the mean of the hot-air experiments with those in cold air , amounting to about T-U of an inch , has been disregarded in the final reduction .
An attempt to test the constancy of the temperature-correction in vacuo , with reference to a suggestion made by Colonel Walker , Supertendent of the Great Indian Survey , who suspects that the coefficient ot expansion of a pendulum in air varies slightly from that in a vacuum , proved a failure .
The pomatum which is used for tightening the different parts of the receiver melted by the heat of the stove , and rendered it impossible to reduce the pressure in the receiver sufficientlyfor the purpose of the experiments .
I. Experiments made in cold air .
II .
Experiments made in hob air .
No. No. of No. TNo .
of of TepeoPressure .
vibrations of TempPressure .
vibrations exp .
rature.per day .
exp .
ratureper day .
I. 47'84 30'052 86oi3'6o ( 70'6 29964 8600272 2 .
47'77 30 ' I2 8603-76 7I ' .
3 29970 86002'64 3 .
4633 30182 86oi3'82z i3 72I 29 958 8600250 4.46 25 29'938 86o4'02 4 .
70'8 29'950 86002o90 5 47'72 30o460 86o0 3-58 5 .
73'8 29960 8600o243 6 .
4791 29'498 860I3'64 .
( 82.5 29'914 85997'84 7 .
4843 29'582 860o 398 2 .
82 5 29'988 85997'10 8 .
4776 29z328 86014'4o 3 .
839 29-960 85995'65 9 .
45'09 30'202 86oI390 4 883 29'941 85992'23 10 .
46.19 30o'oI 860o3'99 5 .
90'8 29'982 85992'73 11 .
47-50 30'0o7 8603-6i 6 .
99 9 29'971 85988'43 The following are the mean results , with their respective difference : Temp. Pressure .
Vibra0 inches .
tions .
A. Experiments in cold air ... ... ... ... ... 4716 ... ... 29'952 ... . .
86oI3'85 B. Experiments in hot air , first set ... ... 71-64 ... ... 29960 ... ... 86002o64 C. Experiments in hot air , second set ... 88o00 ... ... 29 *959 ... ... 85994'00 Resulting differences of temperature and number of vibrations : Temperature .
Vibrations .
A. B =_ 24-48 ... ... II'2I AC = 40'84 ... ... 1985 BC I6'36 ... ... 8'64 494 [ June 17 , Hence we find a correction between 00 11 21 47-I6 and 7I'64 of ==458vibrations 2448 per diem for one 47'16 , , 88-00 , , I98486 , , degree of Fahr7440 84 8| enheit 's scale .
71-64 , , 88o00 -6= '528 J 16-36 Comparing these results with those obtained by General Sabine ( Phil. Trans. 1830 , p. 251 , &amp;c. ) , we find that the pendulums employed by him gave a correction of 0'44 of a vibration per diem for each degree of Fahrenheit between 30 ?
and 60 ?
, a result which agrees well with that found by ourselves between 40 ?
and 70 ?
, the small difference being probably referable to a difference in the composition of the metal of which the pendulums were made .
But a considerable difference appears in the experiments made at the higher temperature .
General Sabine made some experiments , previously to those discussed in the above-mentioned paper , with two different pendulums in a chamber artificially heated to between 80 ?
and 90 ?
, which gave for the correction for each degree of Fahrenheit , respectively for the two pendulums , 0'432 and 0'430 vibrations , corresponding to that part of the thermometer-scale which is included between 45 ?
and 85 ? .
These results are somewhat different from those which are obtained for the scale-reading between 30 ?
and 60Q , and General Sabine points to this difference in the following words *:"In the experiments in the chamber artificially heated , the fluctuations of temperature , in spite of every precaution , were considerable , and rendered the determination of the mean temperature more difficult , and probably less exact than in the natural temperatures ; hence it would be unsafe to conclude in favour of the inference to which these facts would otherwise lead , that the correction at high temperatures is less than at low temperatures , or that the metal expands a smaller proportion of its length for one degree between 85 ?
and 45 ?
than for one degree between 60 ?
and 300 .
" Our own experiments , on the other hand , seem to agree with the general fact that the coefficient of expansion increases with the temperature , and that in a series of experiments a lower range of temperature will give a lower , a higher range a greater value for the expansion for one degree .
Nevertheless the values resulting from our high temperatureexperiments appear decidedly too large to be explained solely by this general behaviour of bodies ; and in our reductions of the pressure-experiments , where the differences of temperature , as will be seen in the following paragraph , are inconsiderable , we have adopted that value for the temperature-correction which results from the experiments between 45 ?
and 70 ?
, viz. 0'458 of a vibration for one degree , a result which not only well agrees with those found by General Sabine , but also appeared to our own considerations the most reliable , for reasons which will appear presently .
Speaking generally of the subject of the temperature-correction , we must admit that our experiments do not tend to remove the difficulties that seem to surround it .
Our experience goes to prove , what the Indian officers , entrusted with the pendulum-experiments and their reduction , have also suspected , that the thermometers fixed to a so-called dummy-bar ( in order to place them in conditions similar to the swinging pendulum ) do not give a true indication of the real temperature of the pendulum .
If this is the case , the differences found by ourselves between the result of the lower and that of the higher range can easily be understood .
Indeed , during the progress of these experiments it has always appeared to us that not only the fluctuations indicated by the thermometers are greater in range than those to which the pendulum itself is subjected , but that they are also more rapid , and that the heavy and substantial pendulum cannot keep time in these changes with the light and delicate thermometers which are not absolutely sealed up into the substance itself .
In our experiments , each of which lasted from one to two hours , a high temperature was usually produced at the beginning , and we attempted to maintain the heat as much as possible by keeping the pendulum-room closely shut on all sides during the progress of the experiment .
The inrush of cold currents can , however , obviously not be wholly prevented , and a steady , more or less considerable fall of the temperature is recorded in each of the experiments beyond 70 ? .
This fall affects , in our opinion , chiefly the thermometers themselves , while probably the pendulum maintains its higher temperature much longer .
Thus we are inclined to think that the mean temperature of the pendulum , if it could by some means be exactly ascertained , might perhaps appear considerably higher than the mean of the thermometer-readings recorded ; and to this circumstance we ascribe it mainly that the high temperature-experiments give too large a correction , for in these experiments a greater difference in the number of vibrations corresponds to an apparently smaller difference in temperature .
The question will , we have reason to hope , find its best solution by the labours of Colonel Walker and Captain Basevi in India , where these gentlemen can avail themselves of a great natural range , which will free the experiments from the doubts and difficulties met by ourselves ; but we cannot conclude this part without reminding experimenters of the words with which , nearly forty years ago , General Sabine concluded the account of his own experiments , and which have gained new force by the shortcomings of our own investigations .
" It seems therefore desirable , for the sake of experiments , which are becoming greatly multiplied , and which are daily increasing in accuracy , that means should be devised of obtaining the rates of pendulums in * Phil. Trans. 1830 , p. 253 .
496 [ June 17 , artificial temperatures , embracing a wider range than the natural temperatures , but capable of being determined with equal accuracy .
" 7 .
There remains now only to give the results of the experiments made for determining the changes in the number of vibrations of our pendulum produced by varying pressures , and hence the correction necessary to reduce experiments made at any pressure to a vacuum .
These results , as given by each separate experiment , are contained in the following Table : TABLE II .
Experimentsfor determining the number of vibrations made by Kater 's invariable pendulum at different pressures .
Full atmospheric pressure .
Temp. Pressure .
Vibrations .
About 25 inches .
Temp. Pressure .
Vibrations .
About 20 inches .
p res .
Temp. Pressure .
Vibrations .
0 inches .
inches .
0 inches .
I. 47d84 30o052 860I3'60 45'82 24-632 86o04'57 47-66 i9'895 86015-87 II .
47'77 30o'I2 86oi3'76 45'54 24'634 860I4'70 47'2I 19'852 86015'99 III .
46-33 30o182 86013'82 46-03 24-620 86014'71 51'04 I9'902 86o016o0 IV .
46'25 29938 86014'02 47'I9 24'599 86o04'49 48*00 i9'9i4 86016'07 V. 47'72 304.60 86013'58 51'oi 24'651 860.1460 46-47 I9'92I 86oi6'io VI .
47'91 29'498 860I3'64 49'84 24-646 860I4'70 49'00 i9'864 86015'94 VII .
48'43 29-582 860I3-98 VIII .
47-76 29'328 860I4'40 , IX .
45'09 30'202 86013'90 X. 46-19 30o'0I 860I3'99 XI .
47'50 30'107 86013-61 About 15 inches .
About Io inches .
Between 7 and 8 inches .
No. of experiPresVibraPresVibraPresVibraeont .
Teip .
sure .
tions .
Temp sure .
tions .
Temp sure .
tions .
meant ; .
TemYlp .
^ Tens .
tiens .
inches .
inches .
0 inches .
I. 43'47 I4'563 8601791 50 ?
75 9'998 86oI9'65 49-37 7'586 86021z61 II .
44'30 14'567 86o0784 45'54 9'868 86019'29 54-11 7'49I 860o2141 III .
50'07 14'532 86018'20 47-13 9'900 86019'43 51'29 7'303 86021'30 IV .
5171 14'680 86017'95 48'24 9'870 860I9'60 500o6 7'554 86021'44 V. 47'09 14'575 86o01800 49'0I 10-015 86019'35 53'42 7'601 86021o19 VI .
46-88 14499 86017'95 51-77 10'064 860I9'51 48'73 7'384 86021o58 VII .
54'33 I4'555 86017'93 Between 5 and 6 inches .
Between 4 and 5 inches .
Between 3 and 4 inches .
No. of experiPresVibraPresVibraPresVibrament .
Temp. sure t. Te tmp .
tions .
I. II .
III .
IV .
V. VI .
o 52I 3 47'86 48-90 5I'30 49'63 54'17 inches .
5'46 i5 420 5'510 5'403 5'390 5'489 86022'10 86022II 86021o97 86021'84 86022'19 86022I15 0o 5'107 51-17 54'o6 50'39 49'84 54'77 inches .
4'241 4'245 4'440 4'530 4.'o7 4'107 4-298 86022'48 86022'71 86022'69 86022o69 86022'37 86022'54 0 5I'34 50'95 50'80 52'09 54-16 53'7 ?
inches .
3'I44 3'i55 3'266 3'204 3'170 3-104 86023'03 86022'90 86022'94 86022'70 86022'71 86022'95 No. of experiment .
e TABLE II .
( continued ) , Between 2 and 3 inches .
Between I and 2 inches .
Below i inch .
No. of experiPresVibraPresVibraPresVibrament .
Temp. r to Teemp .
meant .
Temp. sure .
tions .
emp .
sure .
tions .
Temp. on 0 inches .
0 inch .
y inch .
I. 51i50 2'373 86023-23 60-97 I'393 86023'35 62-76 0'472 86023-26 II .
48-93 2'462 86023'09 59'47 I-4i6 86023-24 60-85 0-43I 86023-47 III .
55-55 2'501 86023-05 58-32 i-4 i 86023'04 60-79 0'444 86023'60 IV .
54'76 2'389 86023-2I 6I'I5 1-430 86023'45 6I-57 0-45I 86023'74 V. 52'38 2-417 86023'08 60'83 I1471 86023'65 62-36 0'425 86023-79 VI .
54I-4 2'451 86023'23 57'58 I'319 86023'I7 60-77 0'389 86023-31 VII ... ... ... ... ... ... ... ... ... ... ... .
49'72 0-427 86023-55 TABLE III .
Mean results of Pressure-experiments .
Mean Mean number Mean Mean number pressure .
of vibrations pressure .
of vibrations inches .
per diem .
inches .
per diem .
I. 0'434 ... 86023'53 VII .
7-486 ... .
8602I'42 II .
1-407 ... . .
86023'32 VIII .
9'953 * ... .
86019'47 III .
2'432 ... ... 86023-15 IX .
14'569 ... 860I7'97 IV .
3'I74 ... .
86022-87 X. 19-891 ... . .
86oi6-00 V. 4'310 ... ... 86022'58 XI .
24'630. .
86014-63 VI .
5'445 ... .
86022'06 XII .
29-95 ; ... ... 86013-85 while Table III .
contains the resulting means for the different sets , as specified in paragraph 4 .
The experiments made at a full atmospheric pressure are the same as those given previously in connexion with the temperature-experiments , but they are here repeated for the sake of comparison .
Their mean temperature being 47 ?
'16 , the whole of the other experiments has been reduced to the same temperature by means of the coefficient adopted in accordance with our preceding statement .
The results as given in Table III .
do not require any special remarks .
It will be seen that the resistance of the air to the motion of a pendulum , as measured by the number of its vibrations , increases very slowly up to 7 or 8 inches of pressure ; a more energetic action is exerted up to about 20 inches , and after that point the resistance increases very slowly up to the full atmospheric pressure .
This behaviour is represented in a more impressive manner on the accompanying curves .
One of them , marked A , shows simply the resulting number of vibrations at the given pressures , which latter form the abscissae , while the former are the ordinates .
The second curve , B , is derived from A , by assuming the whole correction necessary to reduce the pendulum observations made in air to a vacuum as unity , and expressing the correction for intermediate pressures as fractions .
The ordinate representing unity has been divided into forty parts , each representing 0'025 , enabling us to represent the correction to three decimals with great precision .
498 The straight dotted line , C , gives the old correction , and shows best how it differs from the correct one .
m-L___ lb , ; _aI1 lI l lIraIll .
s~~~~~~~2 UJUVVu000 tn 0 r &gt; 0bbbO0Ono0o00o000 00 00 00

