Skip to main content

Full text of "Advanced Practical Physics for Students"

See other formats










r.SSs'iV.i' w.ii*s I ^^H .^ikiatettMsfitti*. 1 ^H = 

WiSii^-'i'^r - 'r'".:'.'.r" 


^ii*---. H .tJI & .-l^lVr^t 


iff, JbdUbU-$**^t 









H. T. FLINT, Ph.D., D.Sc. 







First Published . . . April 1923 
Second Edition, Revised J 9 3 7 



THE course of Practical Physics described in this book 
is based upon that followed in King's College, London, 
by students who have completed their Intermediate 
Course, and who are proceeding to a Pass or Honours Degree. 
This has been extended, and it is hoped that the book will be 
useful to a wider circle of students of Physics than those imme- 
diately concerned with University Examinations. 

A number of well-known Physicists have contributed to the 
development of the King's College course, amongst whom we 
may mention Professors H. A. Wilson, C. G. Barkla, H. S. Allen, 
and W. Wilson, who formerly worked here in the Wheat stone 
Laboratory, and Professor O. W. Richardson, the present occupant 
of the chair. 

The general aim has been to provide with each experiment a 
short theoretical treatment which will enable the student to 
perform the experiment without immediate reference to theoreti- 
cal treatises. To aid this scheme an introductory chapter in the 
Calculus has been included. This chapter is an innovation in a 
book of this type, but it is hoped that the student will find here 
a bridge over that period during which his Physics demands 
more advanced mathematics than his systematic study of that 
subject has yet given him. 

We take this opportunity of expressing our gratitude to 
Professor O. W. Richardson, who has allowed us to make use of 
laboratory manuscripts and results of experiments. We are also 
greatly indebted to our colleagues and to Mr. G. Williamson, who 
have given us many suggestions, and to the Honours students of 


the past session who have supplied us with numerical and graphical 
results. We have been greatly helped by the ready assistance 
on the part of The Cambridge and Paul Scientific Instrument Co., 
Messrs. Elliot Bros., Gambrell, Ltd., Adam Hilger, Ltd., W. G. 
Pye & Co., and the Weston Electric Co., who supplied us with 
the blocks for many of the illustrations. 

B. L. W. 
Wheatstone Laboratory, H. T. F. 

University of London, 
King's College. 
March, ig2j 


IN this revised edition we have removed the misprints and 
errors which occurred in the first. 
We have added to Chapter XXIV on Radioactivity, 
and have included a chapter which contains a miscellaneous 
collection of additional experiments. 

We wish to thank those who have made suggestions for 
additions and who have pointed out errors in the original text. 
The adoption of all the suggestions would have made the book 
unwieldy, but we hope that the additions made will increase its 
value to the student. 

To thank, individually, all students, colleagues, and others to 
whom we are indebted would add greatly to the length of this 
preface. But we feel that we owe our special thanks to Mr. 
Brinkworth, Dr. D. Owen, and Prof. Rankine for suggestions ; 
to Dr. K, G. Emeleus for suggestions and help with the proof ; 
and to Prof, E. V. Appleton, who has added several experiments 
with thermionic valves in the new chapter. 

B. L. W. 
H. T. F. 

February 21, 192J 





























MASS - - - " " " 3 1 


ELASTICITY - - * " - 80 


VISCOSITY ------ 149 




REFLECTION - - - • -258 

REFRACTION - - - " - 2J2 


PHOTOMETRY _ _ - - - 380 

SOUND - - . - - - - 39 2 













APPENDIX ------ 654 

INDEX ------ 655 



The Differential Calculus 

1. Any quantity % which may assume a series of values is called 
a variable quantity or simply a variable, and if its value docs 
not depend on that of any other quantity it is called an 
independent variable. 

On the other hand a quantity y, which bears a particular 
relation to x, assumes values which depend on the values of x, 
and for this reason is called a dependent variable. We may 
have for example : 

y = 2x - 3. 

Here y takes values which depend in a quite definite manner 
on those of x. 

We may also have a dependence defined by the relations : 

y = sin x, y = log x, and y m §•. 

Such expressions as 2% — 3, sin x, log m, etc., are called functions 
of x, and when we say that y is a function of % we mean that 
y depends on the values that % assumes. 

In case we do not specify definitely how y and % arc related 
wc write 

y = /(*)■ 

f(x) denotes any function of %. 

It is often convenient in Physics to show by means of a diagram 
the relation between two variables y and x. For example, a 
record may be required of the atmospheric pressure at various 
times. U Such a record is drawn automatically by a self-recording 
barometer so that it can be seen how the pressure and time are 
related. Here we have as independent variable the time and 
the dependent variable is the barometric pressure. 


In fig. i, the curve represents the relation between x and y 
and the shape depends on the way y and x are connected. 


Fig. i 

If y — 2X — 3, the curve becomes a straight line, and 
if y = sin x, we have the familiar sine curve, fig. 2. 

Fig. 2 

A function is said to be a continuous function of a variable 
when the graph representing it is a curve in which there is no 
sudden change in value of ti\e ordinate at any point. In such a 
curve, if we approach a point where x = a, from left to right, 
we find a certain value for.y, and if we approach the point from 
right to left we find the same value. In fig. 3 we have an example 
of a function which is discontinuous at x = o. If we approach 
the origin from left to right the value of y is very great and 


negative in sign, while in approaching from the right y is very 
large and positive. In nature we are chiefly concerned with 
continuously varying quantities. If a train is at rest at a station 
at a particular instant, and is observed to be moving with a 
velocity of ten miles per hour ten minutes later, it must have 

possessed every possible velocity between zero and ten miles per 
hour during the interval. 

The speed is continuous, and if it depends on the lapse of time 
from the start it is said to be a continuous function of the time. 

We do not contemplate the possibility that the train could 
possess a speed of five miles per hour at one instant and at the 
next without any interval whatever a velocity of six miles per 
hour. If this were possible we should describe the speed as 
discontinuous, because it had no value between five and six. If 
this appeared to be the case we should consider that our powers 
of observation were at fault, and we should describe the motion 
as changing very rapidly between five and six miles per hour ; 
so rapidly that we had failed to detect the lapse of time in which 
the change took place. 

Discontinuous functions are of frequent occurrence in Mathe- 
matics. Consider as an example the case of y = -. 

When x is a very small positive number, let us say — -;, 
y is large and has the value 10 8 . 


On the other hand if x = 6 , y is large in magnitude but 

negative, it equals — io 6 . 

As x passes through the value zero y suddenly leaps from an 
enormously large negative value to a very great positive value, 
and has no value between. 

This is represented in the diagram, fig. 3. The curve has 
two branches : they are the two parts of the rectangular 

xy = 1. 

We shall not be concerned with such functions so we dismiss 
them briefly. It is to be borne in mind that our applications of 
the Calculus are to continuous functions only. The results we 
obtain must not be applied to discontinuous functions without 
closer examination. 

It is important to understand the meaning of the limit of a 

Suppose y depends on x, and that as x approaches the value a, 
y approaches the value b. 

b is called the limit of y as x approaches a, and we write : 

Lim. y ► b 

x > a 

If reference be made to fig. 1, as x approaches the value OM t 
y approaches the value M{9 X and M^ is actually the value 
of y when x = a. 

Cases occur in which the conception of a limit is not so simple. 
If we examine the curve 

J x 
in the neighbourhood of the origin as x — > o, we obtain 
a different value of y according as we begin on the right or left 
of the origin. 

On account of the discontinuity the limit of y as x approaches 
zero is not definite. 

Another case occurs in which x may continue to increase to 
any extent while y continually approaches some particular value. 

We may turn once more to the curve 



As x gets larger and larger, y gets smaller and smaller approaching 
the value zero. 

We may get as near zero as we please by making x larger. 
For example, we may make y as small as one-millionth by 
choosing x — 10 6 . 


This is a very important point in denning a limit. It must 
be possible to get as close to the limiting value as we please by 
choosing x properly, although it may not actually be possible 
to cause y to attain the limit. We have in our example a case 
in point, y is never zero however large x may be, but it is 
possible to make y nearer and nearer zero by increasing x. 

The former definition of the limit of a function is not very 
satisfactory. A limit is accurately defined as follows : 

The limit of a function of x is some number, b, such that as x 
approaches a particular value, a, the difference between b and the 
function may be made as small as we please by taking x sufficiently 
near a. 

2. In describing natural phenomena by means of equations, 
simplifications are often brought about by neglecting certain 
terms in comparison with other more important terms. 
Suppose an equation is obtained which we may write : 

A x + B x + C 2 + D 2 = E x + F 8 . 

The suffix numbers denote the order of importance of the terms ; 
that is to say, 1 denotes that the term is to be regarded as of 
first importance, or it is of the first order of magnitude. The 
2 and 3 denote that the terms are only of second and third 
degrees of importance, they are of the second and third orders. 
If we wish to include terms of the first and second orders we 
omit F 3 , while if only terms of the first order are to be considered 
the equation becomes : 

h x + B x = E x . 

Great care has to be exercised in thus drawing up a scale of 
magnitude, and this leads to a short consideration of infinitesimals. 
Suppose a quantity X is divided into 1000 equal parts, these 
again subdivided in the same way, and so on. We then have a 
series of values : 

X, — 3— 6 ,etc, 
10 3 10 6 

which provides a scale of magnitude. 

If circumstances do not permit of accurate observation of 
quantities less than those of the same order as X we regard 

— r» — / etc., as negligible. 
io 3 io 6 ° ° 

Generally, if / is a small fraction, i.e. small compared with 

unity : 

/X,/*X,/ 3 X, etc. 

are all small compared with X, and are said to be small quantities 


of the first, second, third, etc., orders. If these small quantities 
have zero limits they are called infinitesimals. 

In equations between infinitesimals only the terms of the 
lowest order are to be retained, i.e. the terms of greatest magni- 

This is made clearer by an example which has important 
Physical applications. 

In fig. 4, AB represents the radius of an arc, BP, of a circle 
which subtends an angle at A. 
PN is normal to AB. 

BT is also normal to AB cutting AP produced in T. 
It may be regarded as an axiom that : 

PN < arc PB < BT. 

We shall examine the order of the differences between these 
quantities if be regarded as of the first order of small quantities. 
By expansion of sin and cos in powers of we have : 

03 05 

sin 6 = -, -1 , — , . (-i) 

3l 51 

02 04 

cos = I r + — . — ( 2 / 


If only small quantities of the first order are retained : 

sin = 0. -» . 

cos = 1. / (3/ 

PN = a sin 0. 

PB = a 0. 

i 3 5 ^ 

.\ PB - PN ,= a(0 - sin 0) = a { - - - ( + .. . .1 

This difference is of the third order. 

Thus up to considerations of magnitude of the third order 

PB = PN. 



BT =* a tan 0. 

A 3 2 

= a(0 + -~+ — 5 + - • • )» 

as may be shown by division of the expressions for sin and cos0. 
Thus BT - arc PB = a quantity of the third order of magni- 

Fig. 5 

BN — a — AN = a(i — cos 0) = a quantity of the second 
order of magnitude. 

Thus, if we regard as of the first order and retain only this 
order in our equations we may write : 

BN = o, PN = arc PB = BT. 

and with the exception of BN = o this is true for the case when 
second order quantities are retained. 

Extensive use is made of these relations in Geometrical Optics 
in the first study of reflection and refraction in mirrors and lenses. 

In the case of a mirror, for example (see fig. 5), when the angle 
is small, i.e. when the rays from an object, P, strike the mirror 
at M not far from the pole, O, we establish certain formulas by 
assuming that and N may be regarded as being coincident. 

This is because we do not retain quantities of an order higher 
than 0. Thus NO = o by the foregoing considerations. 

Another important case is the calculation of the order of the 
difference between the sum of two sides of a triangle and the 
base when the base angles are of the first order of small quantities. 


Referring to fig. 6, we have as for fig. 4, 

PA — AN = a quantity of the second order in 0, 
and similarly PA 1 — A X N is of the second order in 1 

.'. PA + PA 1 - (AN + A X N) = a quantity of the second 
order since and 1 are by the data small quantities of the first 

/. PA + PA 1 = AN + A*N = AA 1 
to the first order. 

This result is made use of in the establishment of Fermat's 
Law of the extreme path which plays a fundamental part in 
Optics. (See for example Houstoun's " Treatise on Light," 
p. 17.) 

We consider as a final example the difference between a chord 
and an arc both subtending the same small angle at the centre 
of a circle. 

Thus, again referring to fig. 4, we require the difference between 
chord BP and arc BP. 


arc BP — chord BP = ad — 2a sin — . 


= a 

e - 2( --- ^- + ...) 

_ 1 0* 

2 3! 8" 
= a quantity of the third order. 

We can thus regard the chord and arc as equal up to and including 
quantities of the second order. 

It should be noted that the successive orders are vanishingly 
small with regard to the terms earlier in the scale, e.g. in com- 

aO, bd*, c0 s , 

hf}2 fia 

if is of the first order, --- = — so that as > o 60 2 — > o 

ad a 

infinitely more rapidly than aO, and the same holds for any two 
consecutive terms in the scale. 

The ratio of two quantities of the same order will be a finite 
quantity— not a vanishing or negligible quantity, but the ratio 
of two quantities of differing order (higher order -^ lower order) 
is vanishingly small. 

We are concerned with small variations of this kind in the 
differential Calculus. 

3. The Differential Coefficient 

Let y be a function of x, and suppose x varies by a small 
quantity which we denote by dx. In consequence of this varia- 
tion y will vary a small quantity, say dy. 


The ultimate ratio 4^- when dx becomes very small is called 

dx J 

the differential coefficient of y with respect to x. It is denoted 
by -j-y and written -j- and sometimes denoted simply by Dy. 

In accordance with our notation we may write : 
dy _ Lim. #y 

dx ~ Sx > o dx' 

In general the quantities dy and dx are of the same order of 

magnitude, and -^- is a finite quantity. 

In order to illustrate this, consider the relation : 

y = 2.x — 3. 

Let x become x + Sx, then the new value of y is 

2{x + dx) - 3 

i.e. y + <*y = 2 (* + **) — 3 

= y + 2<S#. 

.*. <5y = 2<5z. 
• & - 2 

Now no matter how small dx becomes, the ratio is always 2, for 
dy is of the same order as dx, and their ratio is finite and equal 
to 2. 

We have a simpler case still in the differential coefficient of 
a constant. 

A constant is a number that does not depend on the variable. 

Thus, if y = A it does not matter how x varies, y still remains 
= A. Thus there is no change dy corresponding to a change dx. 

Hence -j- = o if y is a constant. 

It should be noted that -~ does not mean dy -^ dx. g~ is a short 

. dy 
notation for the operation of finding the ultimate ratio — 

Nevertheless Physicists continually appear to use the coefficient 
as if it meant dy -^ dx, and it is not a rare occurrence to find 
an equation : 

written alternatively dy = x 2 dx. 

This is, in fact, a very convenient mode of expressing the result, 


and it means that dy and dx now no longer retain the same 
significance. The second of these means : 

dy = x 2 dx. 
We have in the equation j- = x 2 an expression of the rate of 

variation of y with respect to a; at a particular point on the 
curve, which represents graphically the relation between y and x. 

The alternative equation means that in the neighbourhood 
of this point we can calculate a small change dy corresponding 
to a small change dx. This point rarely causes difficulty in 
practice, and it is obviously inconvenient to change to and fro 
from d to 8, but to be strictly accurate we must bear the 
distinction in mind. 

The definition of the differential coefficient gives the clue to its 
determination. We will not determine its value for more than 
one or two cases but be content with reference to a table of 
values of the important coefficients. 

The Differential Coefficient for x n where n is any Number. 


y = x*. 

(Sx\ n 
i+ — f 

= x" 

dx _j_ n(n — i) m (toy 

I + "¥ + ^Tr^-i-»^+--- 


fi(n — i) 
= x n + nx*- 1 • dx + - L — — - - x n ~ 2 (dx} 2 + 


= nx»-*dx + n ^~^ • x"- 2 (dx) 2 + . . . 


ty n(n — i) 

.'. v- = nx"* 1 -\ — • x*~ 2 dx + higher powers of dx. 

dx is a quantity which we have called infinitesimal. In the 

next step of finding the limit of ~- we shall suppose dx a quantity 

of the first order of magnitude. It is therefore infinitesimally 
small with regard to the finite quantity nx n ~K 

We thus neglect all quantities of higher order than nx n ~ x and 
have : 

dy Lim. dy 

dx dx >odx nx ' 



Differential Coefficient of sin x. 

y -f. Sy = sin (x + dx) = sin x cos dx + cos x sin <5#. 

We need retain only quantities of the first order on the right. 

Thus : cos dx = I, sin dx — to, by equation (3). 

•*• y + *y = sin x + cos x y - dx. 

.*. dy — cos a; • dx. 

or -3^ = cos x. 

Similarly ■=- cos x = — sin #. 
Differential Coefficient of log x. 

y + <5y = log (X + <5*) = log a/ I + -^j 

= log* + log(i + ^. 

Retaining quantities of first order only : 

dy = dx* — 

"* x 

; *1 = 1 

dx x 

d , t 

or ~-\og X — _. 

dx x 

The same method of treatment can be applied to other cases. 
In the case of a function f(x) we write : 

dy^ _ Lim. f(x + dx) —f(x). 
dx ~~ to > o fa 

The Differential Coefficient of the Sum of two Functions. 

If y = sin x + cos # 

we have -r- = cos # — sin x. 


From the definition it follows that the differential coefficient 
is the sum of the differential coefficients of sin x and cos x. 
In the general case if : 

y=yi+y 2 


where y x and v 2 are any two functions of x : 

dx dx dx 
And similarly if y =y 1 — v 2 

dy _ dyj _ dy 2 . 
dx ~ dx dx 

Differential Coefficient of a Product. 

Let y = y^y 2 

where y x and v 2 are any two functions : 
e.g. we might have : 

y — sinx x # n , 
sin # and #» are two functions of x. 

Suppose that x becomes x + Sx and in consequence y increases 
toy + dy,y 1 toy 1 + dy lt y 2 to v 2 + <*y 2 . 

y + «5y = (yx + «yi) (y« + <5y 2 ) 

Since jy — yiy2 

•*• <?y =^i4y2 + ^1^2 + tyity* 

dy 1 dy 2 is a term of the second order, and the other terms are 
of the first order. 
Thus we need not retain it. 
Dividing throughout by dx. 

&««,& + Si. y,. 

dx yx dx 6x • ra 
Hence in the limit : 

In a product we differentiate one factor at a time, leaving the 
others unchanged, and add all the resulting expressions together. 

This is true for any number of factors, as may be shown in 
the same way. 

Thus, if 

y =ytyzy*y± 

dy dy, , dy 2 , dy 3 , dy* 

& -at ' y w< +y > £** + ™* %* +y ^ y * it- 

e S- y = sin x x % n 


-f- = sin x • nx n ~ x + cos x • *". 



The Differential Coefficient of a Quotient. 

We use the same notation as before and apply the same 

y* + *y* y* v (i + ^) 

(retaining only terms of first order). 


y* 2 

fy y * dx yx dx 
6x ^ y 2 2 

jV — tan x — 

dy __ cos x cos * — sin x( — sin #) 
dx ~~ cos 2 # 

, sin# 
e S> y — tan # = 

J COS # 

= sec z #. 

Differential Coefficient of a Function of a Function. 

The expression : 

y — a sin 9 # -f- b sin x -f c 

in which a, 6, c are constant quantities, is a function of sin #, 
sin x is itself a function of x. 
Thus, y is a function of a function of x. 

We proceed to determine the differential coefficient -j~ in this 

complex case. 

Before attacking the general problem we will consider a 
special case. 

Let y = log sin x. 

and write z = sin x. 


Then y — log z 

and as above : 

dy = — dz, 

J z 

In this step we have regarded y as a function of z. 
But z = sin x. 

.*. dz = COS X'dX. 

<5y = - • cos x ' dx. 






~ sin x 

Note that from 


~ z 

dz we have : 

dy i dz 
dx ~ z dx 


dy i dz 
dx ~~ z dx' 



dz z 

dy dy ^ dz 

dx ~~ dz dx' 

The rule is : 

Differentiate the whole function first 

as if the 

inner function 

(in this case sin x) 

were the independent ' 


and thus obtain 

~, then multiply by the differential coefficient of the inner 

function. . 

This rule, which has been established in a special case, can 
readily be proved generally. 

Let y = F(z) where z =/(*). 

fy = Q .*: 
'dx ~ dz ' dx 


In the limit : ~ = -j- • -j-. 

dx dz dx 

e.g. y = sin 2 * == (sin x) 2 . 

Put z = sin x. 

then y =z\ 

. & = 

dy _ 

dy dz 

~ = 2Z, -5- = cos x. 

dz dx 

, — 2z cos x — 2 sm X cos X. 


It will be convenient at first to introduce z in this way, but 
with practice this intermediate step may be omitted. 

5. The Second Differential Coefficient 

We have seen that when y = sin x t 


-£ = cos x. 


Thus-f- is itself a function of x and will have a differential 


Let -f- be denoted by z. 

dx J 

z = cos x. 

. dz 

. . -j- = — sm x. 

Thus the differential coefficient of -f- is — sin x in this case. 


This is called the second differential coefficient of v, and is written 

Similarly we define higher coefficients -=-^, etc., but we shall 
not be concerned with higher orders than the second. 
As another example consider y =x n . 

~ — nx n ~ x . 

d 2 v 

j+=n(n- i) x»-*, etc. 

6. Applications in Dynamics 

When a particle describes a path under the action of a 
force or set of forces, its position will vary with the time. Thus 
if a stone is thrown vertically upward the position at any instant 
will depend on the time that has elapsed since the moment 
of projection. 

If this distance is measured by y, 

Suppose the time at a particular instant is measured by t and 
the corresponding value of the distance is y. At a small time 


dt later y will become y -\- dy. We may say the average velocity 

. <5y 

is -jf- in this short interval. The smaller we make the interval 

the more accurately will this ratio represent the velocity near 

the position denoted by y. Thus proceeding to the limit in 

which dt ultimately vanishes we have : 

, ., Lim. dy dy 

velocity v — CJ ~ = -£ 

J dt >0 dt dt 

It is often convenient to measure the position of a point by its 
position measured from some point O along the arc it is describing 
(fig. 7). 

Fig. 7. 

In this case -j? is the velocity at P along the arc or the velocity 

in the direction of the tangent at P, where s measures the arc OP. 

In the same way -.— measures the acceleration of a particle 

moving with velocity v., 

dv d 2 s . ds 

But W = df> "^ V = dt 

Thus the acceleration along the arc is measured by the second 
differential coefficient. 

7. A Geometrical Application 

Referring to fig. 1, let Pj denote the point (x- y) and P, 
the point {x •+■ ** • y + <5 3')- 
Then P 2 N = dy and PjN = M^ = 6x. 

The ratio ^ = |^ = tan V^P^ = tan v . 

ox r x JN 

As the limit is approached the line P 2 P t becomes closer and 
closer to the tangent at P x and finally actually coincides with it. 

T , Lim. dy _ dy 
Thus . ^ *■ — r-=tanv 

dx ► o dx dx 


where y now denotes the inclination of the tangent at Pj to the 
axis of x. 

If a curve in the course of its extent is of the character shown 
in fig. 8, the turning points BjBa are called maximum and B 2 B 4 
minimum values of the ordinate. 

Bx is a point such that it possesses the greatest value of y 
for points in its immediate neighbourhood. It is not necessarily 
the greatest ordinate of the curve. 

Fig. 8 

A similar remark applies to the minimum values. At such 
extreme points the tangent is necessarily parallel to the axis of x. 
Thus y = o and tan y also vanishes. 

Hence for maxima and minima : 


This relation is also of great use in Physical problems, as in 
the case of the minimum deviation of a prism or in the theory 
of the formation of the rainbow. 

8. Integration 

The process of Differentiation is to derive from a function 
its differential coefficient or rate of variation with respect to 
the independent variable. 

We have the inverse problem in Integration, where from the 
differential coefficient we have to derive the function., This 
operation is more difficult and cannot be accomplished in every 

case. . 

There are many standard cases which can be readily recognized, 

*e.g. we have seen that for_y = #** 

Y-= nx"- 1 


so that when we are given that 

we remember that y has the value x n . 

But if y has the value x n + A, where A is some constant, 
and hence has a zero differential coefficient we still have : 



•»— 1 

Thus, there is an uncertainty about the value of y to the extent 
of an unknown constant. Such a constant must always be taken 

Fig. 9 

into account in integration and is called the integration constant. 
In Physical problems this constant usually has some special 
value determined by the conditions of the problem. 

We shall approach the Integral Calculus by a Geometrical 

In fig. 9, NM represents a curve, y =/{x). OA and OB are 
two fixed values of x and the area between the ordinates AN 
and BM, the curve NM and the axis of x is divided by drawing 
ordinates P^, P 2 Q 2 , P 3 Q 8 , etc., between A and B, so that 
AP X = dx x , P X P 2 = dx 2> P a P 8 = dx 3J etc. 


All these elements are to be reduced and ultimately vanish. 

Let x v x 2 , # 3 , etc., denote the values^ of the abscissae for points 
lying in the strips AP 1} P^, P 2 P 3 , etc., respectively. 

Of these the ordinate at X 3 , which denotes the point with 
abscissa x z , is a typical example : 

Let the ordinates corresponding to these abscissae be denoted 
by y x , y 2 , y 3 , etc 

We shall suppose that between all the points such as NQ X , 
QiQ2> Q2Q3. etc., * ne curve is continuously increasing or decreas- 
ing. At every point where there is a maximum or minimum, 
as at C, we shall draw an ordinate. The curve is drawn with 
only one such turning point, and this is sufficient for our purpose 
since the argument may be extended to deal with cases where 
several such points occur. 

From the points N, Q lf Q., etc., draw perpendiculars NR 1? 
Q X R 2 , Q 2 R 3 , etc., to the ordinates ¥ x Q lf P 2 Q 2 , P 3 Q 3 , etc., respec- 
tively, and from Q lf Q 2 , Q 3 , etc., draw perpendiculars Q^, Q 2 S 2 , 
Q 3 S 3 , etc., to the ordinates AN, P^, P 2 Q 2 , etc., respectively, 
as shown in fig. 9. 

We have thus two step-like figures which we will call the 
outer and inner stepped figures. 

Now consider the expression y z 8x 3 . 

This lies between Q 3 P 3 x P 2 P 3 and Q 2 P 2 x P 2 P 3 . 

It thus represents an area intermediate between that of the 
rectangle P 2 P 3 Q 3 S 3 and P 2 P 3 R 3 Q 2 . 

In the same way y x fa x lies between APiQ^j and AP^N. 

Thus, if we consider the sum : 

y 1 6x 1 -\-y i 6x 2 + y z 8x 3 + . . . +y n fi%n = tydx, 

where y n represents an ordinate in the last strip which terminates 
at B and has a width 8x n , we know that it lies between the 
outer and inner stepped figures. 

We have to inquire into the value of the sum when we make 
the number of divisions very large or, what is the same thing, 
the values d$ lt 6x 2 , 8x 3 , etc., very small. 

In performing this summation we meet with the difficulty 
that although each term becomes small, the total number in- 
creases, and the question arises as to whether the sum remains 
finite under these circumstances. It seems almost axiomatic that 
the sum will in the limit prove to be the area between the 
curve, the extreme ordinates and the #-axis. 

However small the values, 8x, become the sum concerned 
always lies between the outer and inner stepped figures. The 
difference between these areas is the sum of the rectangles : 

SjQ.RiNi, S.QtR.Qp etc. 


There will always be one strip which has a width greater than 
that of any of the others, or at any rate its width will not be 
less than that of any other, though of course it is possible to 
have another strip of equal width. Let this be the strip lying 
between P r and P r+1 , and draw the two rectangles as before, 
viz. P r P r+1 Q r+1 S r+1 and P r P r+1 R r+1 Q r l 

In our figure the greatest ordinate is CD. Continue P r +1 Q r , , 
to T 1 and P r S,+i to T so that P r+1 T* = P r T = CD. 

Now all the small rectangles S^, S 2 R 2 , S 3 R 3 , etc., lying 
between AN and the maximum ordinate at C could be removed 
and placed within the rectangle TP r+1 . Thus the difference 
between the two stepped figures lying on the left of C cannot 
be greater than the rectangle TP r+1 . 

As the widths 6x lt 8x 2 , dx z , etc., are diminished the area 
TP r+1 diminishes and vanishes in the limit. 

Thus in the limit when the widths of the strips become infinitely 
small, the difference between the two stepped areas also diminishes 

In other words, the two areas become equal in the limit, and 
each is then equal to the area between the curve, the ordinate, 
AN, the maximum ordinate and the #-axis. 

The same process may be applied to the area between the 
maximum ordinate, BM, the curve and #-axis. Thus adding 
the two parts together we find that the total area is equal to 
the sum : 

in which 8x tends to zero. 

If there occur any finite number of maxima or minima in the 
curve, it may be divided at each and the same process applied 
to each partial area. 

When the several areas are summed the result applies to the 
whole curve. 

ydx and it is called the integral 

of y with respect to x. 

a and b denote the values OA and OB, and since these are definite 
abscissae the integral is called a definite integral. 

In some cases when we do not fix the limits we write simply 

J ydx and the integral is called indefinite. The lower limit 

is then tacitly assumed to be some convenient starting point, 
while the upper limit is any variable point which we denote 
by the variable x. 


9. Connection between Differentiation and Integration 

If we consider the area included between the limits A and X 
where X is any point with co-ordinate X, the magnitude of the 
area depends on the positions of the points A and X, and the 
form of the curve. The form of the curve is fixed by its equation, 
and when X is changed the area will vary by an amount depending 
on the change in X. 

The area thus depends on the value of X at which the summa- 
tion is stopped. If the student feels any difficulty about this 
point he should draw a semicircle and choose the diameter as 
x— axis. Choose one end of the diameter for the point, O, 
and determine the area up to an ordinate drawn at a point, X, 
taken on the diameter. If OX is denoted by x it will be found 
that the area can be expressed as a function of x. 

Thus we may write : 

fydx = a function of X, say, /(X) 
Write : 

= jydx. 

J a 

We are about to show what is the relation between A and v. 

Proceed a step farther with the integration, up to the point, 
Xi. Denote XX 1 by <5X. 

The area will alter by the amount, XX^Y, where YY 1 
means the arc of the curve between Y and Y 1 . 

Write this change = dA. 

Then the new area is 

A + dA 

= jydx. 

= area up to XY + XX^Y. 
But A = area up to XY. 

/. 8A = XX^Y. 

= Y x <5X in the limit, 

where Y is an ordinate in the strip, XX 1 , and in the limit we 
shall suppose the strip to possess a vanishingly small width. 

_. T . dA dA 

We have used X and Y to denote the variables for convenience, 
but we have not specified any particular values for them ; they 


denote any abscissa and corresponding ordinate so that we may 
just as well write : 


y ~ dx 

Thus y is the differential coefficient of the integral which has 
been denoted by A. 

The problem thus resolves itself into determining A if we 
know its differential coefficient. 

The occurrence of the arbitrary constant is seen to be connected 
with the choice of the starting point. 

We may summarize this by stating that if : 

= lydx 

then y =^ 

J dx 

There is no general rule for passing from y to A, i.e. from a 
function to its integral. 

We can only perform the operation by recognizing a standard 
form, and many devices have to be studied for the reduction 
of forms not directly recognizable, to more familiar ones. 

For these the student is referred to textbooks on the Integral 

Reference should be made to tables in Mathematical textbooks 
for the important standard forms. 

10. Evaluation of a Definite Integral 

Suppose that A has been determined in this way, and that it is 
now expressed as a function of x. 

Then A = lydx = f(x) + constant. 

= /(*)+B(say) 

B is the constant which it is necessary to add for the reason 
explained above. 

B depends on the arbitrary starting point so that there is 
some connexion between B and a. 

Write x = b, i.e. perform the integration up to a fixed point b. 

Then fydx = /(&) + B. 

If we write x = awe have lydx. 

This must vanish, for it means the calculation of an integral 


of no extent, or referring to our illustration it indicates an area 
of no width. 

.-. fydx=f(a) +B = o. 

/. B = -/(a). 
This is the relation between B and a. 

Hence : fydx = f(b) - f{a). 

The rule is therefore : Find the function of which y is the 
differential coefficient and express it as a function of x. Substitute 
in it the values of the upper and lower limits and subtract the results 
so obtained. 


# 2 is the differential coefficient of 



10 IQ 3 j% 

x*dx= ^- = 333- 



* sin xdx. 

sin x is the differential coefficient of — cos x. 


.-. P sin xdx = [" — cos x\ = - cos ^ - ( — cos o) = 1. 

This measures the area of the part of the sine curve from the 

origin up to its maximum ordinate at -, lying above the x axis. 

11. An important application of the Integral Calculus is to 
the calculation of moments of inertia. (Chapter II.) 

A body is considered as made up of a large number of small 
masses 8m. Let r denote the shortest distance of dm from an 
axis. Then the product r 2 8m is called the moment of inertia 
of the particle 8m about the axis. If all the particles of the 
body are taken into consideration and we make the summation 
Er 2 <5m for the whole body, we obtain the moment of inertia of 

the body. Thus Sr 2 <5w or IrHm is the moment of inertia of 
the body about the axis. 


If we have an area A divided into small elements, da, each 
of which has a corresponding distance r from some axis, we require 

sometimes the value fr 2 da, and this may be called the 

moment of inertia (shortly M. of I.) of the area about the axis. 
Such an expression occurs in the treatment of bending of beams, 
and will be employed in the calculation of Young's Modulus 
by bending. 

As an example of the calculation of a M. of I., consider the 
case of a thin rod of total mass M and length zl. Let its mass 
per unit length be m, so that M = ilm. 

Assume that the axis is through the e.g. and perpendicular to 
the rod (fig. 10). 



- — x— \ 


1 * 

Fig. io 

The form of section is of no consequence, but the ends at A 
and B are perpendicular to the length. 

The element of length dx at distance x from GZ will have a 
mass mdx and its moment of inertia about GZ is mix • x z . 

Thus the total M. of I. = Jmx 2 dx the limits indicating 

between what limits x extends. 

Now CmxHx = [y^] =| ml* = *m(2t)l 2 = —.... (4) 

The application of integration to the calculation of moments 
of inertia is extended in Chapter II. 

12. Oscillatory Motion 

.Many experiments make use of the fact that in a number 
of cases bodies slightly displaced from their position of equilibrium 
perform periodic isochronous vibrations about that position. 
Examples of this occur in the case of the simple pendulum, in 
cases of torsional oscillations, and movements of galvanometer 


We usually require the complete period of the oscillation in 
making calculations. 

Such motion is best and most simply treated by means of the 
Differential Calculus. 

A simple harmonic vibration is by definition one in which 
the body moves so thajt it is under the action of a force tending 
to restore it to the position of equilibrium, the magnitude of 
the force being proportional to the measure of its displacement. 

Thus, if a point is moving to and fro in a straight line 

about a position O, and is performing S.H.M. it is always under 

a force directed towards O and proportional to thd distance 

OP ( = x). 

d 2 x 
But the acceleration is -j^' and if we write,: 

force = — kx 
where k is constant and the negative sign denotes the direction 
of the force, we have by Newton's Second Law of Motion 

d 2 x . 

m M = -**" 
y It is usual to write this in the form : 


where * 2 = — ■• 
r m 

This is an example of a differential equation. 

We do not here consider any series of arguments leading 
logically to the solution of this equation, we merely state the 
solution and verify the truth of the statement. 

This equation and another slightly more complicated are so 
important in Physics at an early stage that the solutions should 
be remembered and the student prepared to apply them with 

Consider x = A sin (j>t + a) (6) 

A and a are arbitrary constants not occurring in the equation. 
They occur for the same reason that B occurred in § 10 ; in fact, 
we are proceeding from a differential coefficient to the function 
from which it is derived and so are performing an integration 
when solving the above equation. It should be remembered 
also that in the complete solution of a differential equation 
there must occur the same number of arbitrary constants as the 
number of the order of the highest differential coefficient in the 

In the present case the highest order is 2, and so we have A 
and a. 

■[£•+#* -o. (5) 



d 2 x 

57 i - = -^ 2 Asin(^ + a) 

Now ~£ = pA cos (pt+«), ^ 

= —p 2 x. 
Hence the value chosen for x satisfies the equation. 
Since x = A sin {pt + a) we shall have the same value of x 
occumng T sees, later when a complete period has elapsed. 
/. x = A sin [p(t + T) + a] = A sin (pt + a). 
.\ sin (pt + a + pT) = sin (£* + a). 
This is true if pT = 2*, 47c, etc. 

Thus the first recurrence of the value of x is after an interval 
T = 2*/p, and this is the period of the S.H.M. described bv 
our equation. 

We note that T = 2tu -f- square root of the coefficient of x in 
the reduced equation / 7 \ 

O ^9 

Fig. 11 

Thus in considering any problem in which T is required we 
have only to write down the equation of motion and but it in 
this form ; we can then write down the value of T immediately 

We consider the case of the simple pendulum (fig. 11). 

The displacement is measured by the angle 6. 

The force along the tangent to the circle described by P is 

?fx Sm - e ? T ™ 8 ° if 6 is smaU - Note that the m °tion is only 
b.H M. if 6 is small, since the force is only then proportional 
to the displacement. 

The acceleration in this direction is ~ 

dt 2 

where s = arc OP. 

or since s = Id, it is / 

dt 2 


The more complicated case occurs when the effect of friction 
has to be included. The force of friction depends on the sum 
of a number of terms proportional to the powers of the velocities. 
In slow motions, such as those occurring in the movements of 
a galvanometer needle, the term of paramount importance is 
that containing the first power of the velocity. As this discussion 
is for the purpose of considering oscillations such as occur in 
galvanometers we consider this as a typical example. 

The position of the needle or coil is defined by its angular 
displacement from a normal position. 

For example, let OH denote the direction of a magnetic field 
in which the needle normally sets. 


A 1 

"^J 3 


B - — H 

Fig. 12 

In the new position, A 1 !* 1 , let measure the angle BOB 1 . 

If is small, the restoring couple due to the field H is 2mUl sin 
= 2mMd, where / = half the length of the magnet. If the 
magnetic moment be M this couple may be written MHO. 
m denotes the pole strength. If I denotes the M. of L of the 
needle, neglecting friction we have : 

!»?--»=• <9) 

and the motion is simple harmonic. 

We assume in accordance with what has been stated above 
that there is a frictional force proportional to the angular velocity 

4i and write the force = c -& where c is constant. 
at at 

The equation then becomes : 

T d 2 d dd __„ , v 

and may be simplified by dividing throughout by I and changing 
the notation of the constants^ 


It will now be written : 

dH , „ dd , , a . . 

— + K Jt+ n*6 = o. .. (ii) 

The derivation of this result has been made by reference to a 
magnetic needle oscillating in a magnetic field, let us say, supported 
by a thread of negligible torsion. 

The same result is obtained for such oscillations as occur in 
the case of bars vibrating at the end of stretched strings under 
torsion or for oscillations in the electrical discharge of a condenser. 

We shall denote ~rr by 0" and ^ by 0*. 

Again, we do not attempt to deduce the solution of the equation, 

we merely verify a stated result. 

The solution of such equations can be obtained by writing 

6 = Ae mt 

for on using this value of the equation is satisfied provided 

m* + Km + n 2 = o. 

This is evident if we substitute = m 2 Ae mt , = mAe mt . 

„ — K ± VK 2 — An* 

Hence m = 3— = m x or w 2 , say. 

Thus the complete solution is : = Ae m i* + ~Be m 2*. 
A and B are the two constants necessary in the complete solution. 
If we substitute the values of m x and m 2 we find : 

e = e iAe 4 +Be * 1 (12) 

K 2 
When — > n* the indices of the exponentials in the bracket 

are real and is not periodic. merely changes in value with t 
according to the exponential law. Of course K and n depend 
on the particular problem considered. K measures the friction 
and n depends on the nature of the restoring force. If friction 

were not present the body would vibrate with period — »so 

that this may be said to be the natural period of vibration if 
there is no friction. 

The case with which we are concerned is when the motion is 
oscillatory, and this requires that the indices of the exponentials 
within the bracket should be imaginary, 

i.e. £1 <n i m 


Write * = V — 1, we then have for the terms in the bracket * 


By altering the constants we can put this in the i orm : 

C sin 
or more simply : 


K 2 

.*+ Dcos 





Esin(^J« a - — -* + <x) 


Thus the solution is now to be written : 

6 = Ee ^sin [yjn* - — • t+ a) 

where E and a are the constants. \ 

The student may, if he wishes, take 6 as given by this value, 

substitute in the equation and verify that this satisfies it. 
This form of the solution shows the similarity with the last 

case in which we had 

== A sin {pt -f a). 


The amplitude is now Ee a instead of A. It thus varies 

— K£ 

with the time, and since K is positive, the value of e a is less 
than unity, and the amplitude diminishes with the time. The 
motion is said to be damped, and the damping depends upon K. 
does not maintain the same amplitude during the motion, 
as in the undamped case. The curve illustrating the motion is 
drawn in fig. 13. 

B = Ee a where t has 

t — 

Fig. 13 
When t = o, O = E sin a, 

and when t = - -r- \\ n 2 — -— - — a, 
2 \ 4 

this particular value. 

Other pairs of values (0, t) may be obtained in the same way, 
and the curve drawn as above. 

It will be noted that the curve crosses the 2-axis at the 
points AxA 2 . . . and that these points are equidistant. 

The interval k x k z is called the period of the motion. 


It will be noticed that 6 is zero when 

sin Ujn »- — * + a) 

For simplicity write q =» J n * _ J£ a 

The n sin (qt+ a) =o if' 

j* + a = o, n, 2tt, 3tt, etc. 

that is when t = - - n ~ a . 2 * ~ a » 3* - « ^ . 

?' <1 ' q q ' * 

On the curve these times correspond to the points A, A v 
a 2 , A3, etc. 

In the case of point A x the oscillator is at the zero position, 
but is moving in a direction opposite to that at A or A 2 . 

The difference between the above values of t at A 2 and A 3 is 

3 TC ~ « _ " — a _ 2n 

? q q' 
Thus the period is L« _ j^2 . . . , / x >) 

Had there been no damping, the period would have been —. 

We shall call these T x and T respectively. 
At such points as B,, B 2 , B 8 , etc., the value of sin (qt + a ) 
is numerically equal to unity, and the amplitude is then measured 

by Ee 2 with the appropriate value of t. 

At B 2 the value of t is^- a and at B a half a period later the 

value is — - a + -T, = ^- - a. 
2? ^2 x 2.q 

The corresponding values of sin (qt + a) are +1 and -1 • this 

means of course that the displacement is on the opposite side 

of the mean position in the second case. 

Thus SA =tf +-^ 

X>2 < -'2 

In the same way 

B 2 C 2 JSIi B3C. 

B£~r e * = B^ = etc (15) 

Thus the ratios of successive maximum displacements are 

The value-— 1 is denoted by x which is called the logarithmic 




The measurement of the length of a body may be made by 
using one of the usual vernier or micrometer devices, such as the 
vernier calliper, the micrometer screw gauge, or the spherometer. 

The measurement of small objects may also be carried out by 
use of the travelling microscope or the micrometer microscope. 

In using either of these instruments care must be taken, when 
viewing the image of the ends of the object, that this image is 
in the same plane as the cross-hair or the small scale in the 
eyepiece, otherwise parallax errors may be introduced. The 
image seen should not show any relative movement with the 
cross-hairs when the eye is moved across the field of view. 

When focussing the cross-hair, as a preliminary adjustment, 
the eye should be unstrained. The microscope is turned to a 
bright distant object and the adjustment of the eyepiece should 
be made so that the distant object viewed by the one eye is in 
focus when the cross-hair as viewed by the other eye is also 
clearly focussed. 

The Comparator 

When a length exceeds a few centimetres, the travelling 
microscope or micrometer microscope are not used individually 
as measurers, but are replaced by an arrangement of two such 
instruments arranged at a variable distance apart on a fixed 
graduated bed. Each microscope may be moved in the usual 
manner in a direction parallel to the length of the bed. Each 
microscope is provided with a scale or fine cross-hair in 
the focal plane of the eyepiece. Each eyepiece is adjusted 
so that the scale or cross-hair is in sharp focus for normal 
vision, and is replaced in the carriage. 

An object to be measured is fastened rigidly along the bed, 
in the groove provided for it. The image of one end of the object 
as seen by the first microscope is brought into coincidence with 
the image of the intersection of the cross-hairs and the second 
microscope is moved until the image of the other end on the 
object coincides with cross-hair intersection in that microscope. 

Thus the images of the two ends of the object as seen by the 



two microscopes normal to its length are formed at the inter- 
section of the two pairs of cross-hairs. 

The object is removed and a standard rule substituted. The 
readings of the standard scale seen opposite the two cross-hairs 
obviously enables the length of the object to be ascertained by 

As an example of this Substitution method we will consider the 
experimental details of the following experiment. 

The Comparison of the Yard and the Metre 

Standard rules, engraved as finely as possible with inch and 
centimetre graduations, and of lengths one yard and one metre, 
are employed in this experiment. 

Place the yard rule on the bed of the comparator and focus 
the one microscope on a scale division near one end of the yard, 
moving the latter and the microscope until this is accomplished. 
Then move the second microscope until a scale division near 
the other end of the rod is sharply focussed : the number of 
inches included between the two being noted — 36 if the scale is 
sufficiently well graduated to allow of this. Then, taking care 
not to upset the arrangement of the microscopes in any way, 
remove the yard scale and substitute the metre, so that the 
graduations are in good focus. Move the scale so that the 
image of a division near one end is in coincidence with the cross- 
hair intersection in the one microscope. Under these circum- 
stances the second microscope will not be opposite a division. 
Coincidence of the cross-hair and the image of a scale division 
is brought about by a movement of the microscope, parallel to 
the length of the bed, which is measured on the vernier scale 
attached to it. 

The size of the gap between the two cross-hairs ± the move- 
ment of the microscope is then read off in cms. 

Care is taken in noting the movement of the microscope to 
see exactly the unit used in these graduations. In this way we 
obtain two measurements, one in each system, for the same 
distance and may calculate the number of inches to the metre, 
or cms. to the yard. 

The vernier scale movement on the microscope is often replaced 
by a micrometer screw capable of a much shorter range of move- 
ment. With such a screw traverse, the movement of the micro- 
scope may be readily measured to «ooi of a cm. Using sucli a 
comparator the values given below were obtained : 

(1) z yard = 91-5 cms. + -008 inch. 
/. 1 metre s= 39/331 inches. 


(2) 50 cms. = 19*6895 inches, 
i.e. 1 metre = 39*379 inches. 

(3) 1 metre = 39*375 inches + -002 inch 

= 39'377 inches 
mean value 

1 metre = 39*362 inches. 


The Planimeter 

The estimation of the area of a plane figure may be carried out 
by one of the many geometrical methods or by the use of a 
planimeter, an instrument designed to measure such areas 
directly. / 

Of this class of instrument the Amsler planimeter is generally 

Fig. 14 

It consists, essentially, of two arms AC and EB, hinged at A, 
fig. 14; AC is of fixed length and is provided with a needle 
point loaded above by a small weight as shown in the diagram. 
The second arm may be varied in length by sliding that portion 
of it which carries the tracer, B, into the slot provided in the 
other half, EA. By means of fine adjustment, S, the length BE 
may be set accurately at any division along the graduated face 
of BA. 

In addition to the tracer, B, this arm is provided with a small 
wheel to which is attached a graduated drum which moves past 
a fixed vernier scale, V. By means of the graduated drum and 
vernier, the rotation of the wheel may be measured to Tff ta of 
a complete turn. The axle of the wheel and drum is arranged 
parallel, to the length of EB and is provided with a worm gear, 
which moves a horizontal indicator, D, one division per revolution. 

When placed on a plane the instrument is supported at three 
points : the needle point C, the point of the tracer B, and the 
point of contact of the wheel with the plane. 


To measure an area the needle C is placed at a point outside 
it, such that B may be moved round the boundary of the area. 

For the measurement of large areas this will be impossible, 
but the details for such a case will be seen later. Starting at 
any point on the boundary of the area, the tracer is moved 
carefully over its contour until it is finally in the starting position. 
During this operation the wheel will have rotated in general a 
definite number of revolutions plus a measurable fraction of a 
revolution. From this observation, the area of the figure may 
be calculated. 

The method of calculating the area will be best understood 
by first considering the theory of the instrument. 

— X 

Fig. 15 

When the needle C is fixed in the plane of the area to be 
measured, any movement of B along the boundary of the figure 
will result in a movement of A along the arc of a circle with C as 
centre and radius CA = a cms. (see fig. 15). Further, the 
wheel, W, will roll a distance equal to the total displacement 
when the movement of EB is at right angles to its length : 
movement parallel to the length causes no rotation, the forces 
acting on the wheel due to contact with the plane, under these 
latter circumstances only produces a couple tending to move 
the axle parallel to its length about the pivots. 

So for any intermediate form of displacement, the rotation 
produced in the wheel will correspond to the component of the 
displacement at right angles to the length EB, and if n revolutions 
occur, 2ttwR will measure this normal displacement, R being the 
radius of the wheel. 

Now it will be shown below that the area £0 be measured is 
equal to 2imR x AB : since the distance of the wheel from A 



does not occur in this expression we must first show that this 
distance has no effect on the number of revolutions the wheel 

In fig. 15, let B and B 1 be two positions of the tracer a very 
small distance apart on the boundary of the area to be measured, 
B B 1 K ; A and A 1 being the corresponding positions of the hinge. 

Let AB = b, and suppose that the centre of the wheel is at P 
(fig. 16). Draw AN 1 and PN normal to AfB' 1 from A and P and 
let AN 1 = ds : PN = ds' : AP = c. 

AL being parallel to A X B*, let the angle LAP = dtp. 

Fig. 16 

Since BB 1 is a small distance, ds, ds', and d<p are also small. 

Now PN = NL + LP 

or _ ds' = ds -f c- dq>. ' 

The distance moved by P as B traces the boundary of the 
area BB X K is S ds' 

or S ds' = S ds + S cd<p. 

B finally returns to the starting point, and therefore 2 dtp = o 
i.e. S ds' = Sis. 

Thus a wheel placed at A would indicate the same movement 
as the one at P, or at any other point along EB as could be shown 
by the same process as above. The position oj the wheel on the 
arm AB does not, therefore, affect the reading of the instrument. 

To show that the area of a plane figure which does not include 
the needle point C is equal to (2imR)b let us refer the figures 
to rectangular axes with C as origin, as in fig. 15 • 

Let CAB be one position of the planimeter and CA X B X a second 
position such that BB 1 is a small displacement of the tracer. 
Let the area to be measured be BB 1 K. 

Referred to these axes let x, y be the co-ordinates of B. 


The area BB*K may be conveniently referred to polar co- 
ordinate r and 0, i.e. let CB be r, /_ BCX = then if /_ B X CB 
= dd, the area B*CB = \r • rdd = \rHB, i.e. 

S \rW = area of BB^ . (i) 

For as the radius vector moves round the figure on the boundary 
remote from O, the small area contains, in turn, each element 
of the area to be determined, + the external triangle from the 
boundary of the figure on the side near to O ; this latter area 
is deducted from the sum as the radius vector travels along this 
near boundary, for here 66 is negative. 

Now x = r cos y = r sin 

dx = — r sin • dd -f cos dr, dy = r cos • do + sin • dr 

whence xdy — ydx = r* dd. 

Or, the area of the figure, £ S rHB from (i) 

= \ S (xdy — ydx) (2) 

Let /__ ACX = a ; A*CA = dx ; AN X A J = 90 ; CA = CA 1 
= a ; AB = A^ 1 = b. 

The co-ordinates x and y of the point B may be expressed as 
under : 

x = a cos a -f- b cos q>, y = a sin a + b sin <p. 

.*. dx = — a sin a dx — 6 sin 9? dg? d[y = a cos a da + & cos yip. 
.*. x dy — j/d# = (a cos a + 6 cos 93) (a cos a-da + 6 cos tp-dcp) 
-f (a sin a +b sin 9?) (« sin a da -f& sin 9>-ig> 

= a 2 da 4- & 2 dg? + ab cos (a — <p) d (a + 9) (3) 

But (a + <p) = 2 a — (a — 9?) or d (a + <p) — 2 dec — d (v. — <p). 
,\ a& cos (a — ?>)*i (a + 9?) = a& cos (a — 9?) {2 da — d (a — ?>)} 

— 2 ab cos (a — • 9>) da — ab cos 
(a — 9?)-d (a — <p) 

AN 1 = ds = AA 1 cos A X AN == a • d acos (a — 9?). 
so that 

aft cos (a — <p)-d (a + <p) = 2b' ds — ab cos (a — q>) • d (a — 9?), 
and equation (3) becomes 
* • dy — y • d# = a 2 dat.+b 2 d<p + 2b'ds — ab cos (a — ?>)• d (a — ?>). 

Now, when the trace moves round the curve, Sdy = o, Sda = o 
for the planimeter returns to the exact position of starting : 
also we have then 

2 cos (a — 9) • d (a — 95) =0, for S^cos (a — <p) • d (a — 9?) 
« £sin ( a - g>) J 


But limits (i) and (2) are identical so that this = o (see page 22), 
i.e. S (x-dy — ydx) = E zbds = ibYds. 

Now Sis = 27twR, as already shown, and we saw in equation 
(2) above 

Z(xdy — ydx) = 2 x area enclosed in BB 1 K, 
i.e. 2 area of figure = 2& 2tcwR. 

/. Area of figure BB X K = b • 27mR. 

Thus to measure any area sufficiently small, C is fixed in the 
paper at a point outside the area and B is taken round the 
boundary. The rotation of the wheel is measured, as is the 
distance from the hinge to B, and the area is thus equal to the 
product of BA and 27mR. » 

To carry out the calculation, R and AB must be measured. 
The distance from the tracer point to the hinge may be estimated 
by holding the instrument parallel to a scale or squared paper, 
and estimating as nearly as possible the position of the axis of 
the hinge on the scale. The difficulty is of course in estimating 
the true position of this axis. 

The value of R may be obtained by measurement with a screw 
gauge, which must be used with great care, as the edge of the 
wheel is easily damaged by screwing up the gauge unduly. 

Another and safer way of finding R is to note drum-reading 
on the wheel, and move the wheel along a straight line ruled on 
paper, until the wheel has made several complete revolutions. 
The distance moved and the number of revolutions enable the 
value of 2wR to be measured directly. 

The two measurements described above cannot be made with 
very great precision, but are performed much more accurately 
in the construction of the instrument. The graduations on the 
arm BA, ' 100 d cm,' etc., are made in the construction, and 
signify that when this graduation is adjusted to the fixed mark, 
one revolution of the wheel corresponds to an area of 100 sq. cms. 
Thus, if the graduations are to be trusted, i.e. unless the instru- 
ment has been subjected to rough handling, the area of the 
figure is equal to«x (the number on the graduated arm opposite 
the fixed mark. 

The Case when the Needle is Inside the Figure. 

Now let us consider the case when large areas are to be 
measured, e.g. BEjE^ . . . En, fig. 17. The needle support 
is fixed at a central point C, and the tracer may then be made to 
trace the boundary line of the area. 

But in this case it will be seen that 2nriR • b does not give the 
true value of the area of the figure ; for in this case ~Ldq> = 2w 
and is not zero. 


Take a point B on the boundary, such that BA is at right 
angles to CW, the line joining C to the centre of the wheel, W ; 
and draw a circle with C as centre and CB as radius, shown in 
the broken line in fig. 17. Then if B is moved round this circle 
the wheel W will move round a second circle of radius CW ; the 
relative positions of the two arms remaining constant so that 
the axle of wheel is always at right angle to the radius CW. Thus, 
while B is moved round the broken-line circle, W is moving 
always parallel to its axis around the second circle. So that 
during the whole revolution the wheel will not rotate. That is 
to say, the circle BFjF^ etc., so drawn, is such that when traced by 







B 9 

Fig. 17 

the tracer, the wheel indicates zero movement. The area of the 
circle is u(CB) 2 and is quite definite in size, depending on the 
setting of the graduated arm. This is sometimes called the zero 
circle, or datum circle. 

If now we consider our area EjE^a . . . E 1]L and commence with 
the pointer at B : passing from B to E 2 via E x , B is moved 
outside the circle. W will move towards C, and a definite 
rotation in one direction is made by the wheel. We could 
bring the tracer back to B along the path E^B without altering 
the reading of W. This reading corresponds, in the way previously 
considered, to the area EaFBE!. 

However, having reached E 2 , continue along the boundary 


E2E3E4. To do this the tracer moves inside the zero circle and 
W will therefore move outside its circle, i.e. in the opposite 
direction to the previous movement. 

Similar movements occur round the figure. The planimeter 
therefore adds algebraically the area of the curve outside the 
zero circle. Having carefully noted the direction of rotation 
of W when B traverses such a part as E 2 , we can tell from the 
final reading of the dial D whether the figure is of less or greater 
area of the zero circle. Suppose n revolutions of the wheel are 
indicated and the arm is set at/the ' 100 □ cm. ' mark. If the 
indication of the n revolutions is in the same direction as the 
indication of the wheel when moving outwards, e.g. along BEiE 2 , 
the area^of the figure is 

100 • n + nCB 2 sq. cms. if CB is measured in cms. 


Fig. 18 

The same difficulty as before is met with in finding CB. The 
instrument may be set along two lines drawn at right angles so 
that the point of contact of the wheel is at the intersection of 
the lines, and B and C are each on one line, fig. 18 ; CB is measured 
directly or WB and WC are measured and CB calculated. How- 
ever, the value of this zero circle is inscribed on a second face of 
the arm BA. Adjacent to the '100 □ cm.' mark is a number 
which gives a value of the area of the zero circle, not usually in 
sq. cms., but in revolutions. 

Thus, if in the case taken above, there are n revolutions in- 
dicated and the second scale gives m, as the equivalent area of 
the zero circle, the area of the figure is 

100 (n + m) sq. cms. 

To become acquainted with the instrument and familiar with 
the method of using, draw several small regular figures, calculate 
the areas, and then find them, using the planimeter at different 
graduated scale-settings. 

Draw a circle of known radius, calculate the area. Move the 
tracer round the circumference when the movable arm is set 
at various graduations. Note the number of revolutions in each 
case. From the calculated area and the number of revolutions 


observed find the area corresponding to one revolution, and 
so check the graduations. Repeat this process with several 
measured areas, and obtain a calibration of the instrument. 

Measure b and R and again calculate a value for the area 
corresponding to one revolution. This is not so accurate as 
the above method, but shows the value of the construction as 

Draw irregular figures on squared paper and find the areas 
by adding the squares. Compare these results with those 
obtained by the planimeter as calibrated. 

Measure the radius for the zero circle at each setting, and 
check the value of the graduated scale, by calculating the area 
and dividing by 2nRb. The quotient should agree with the 
graduations on the scale. 

The instrument may subsequently be used for area measure- 
ments when required, making use of the calibration if errors 
are found by these experiments. 


When the boundary of the figure does not cut the zero circle 
the process is identical ; for, suppose ABCD be such a figure, 
and the zero circle is completely inside the figure as shown. 

The process is as before. Start at any point A, and, with the 
fixed point inside the figure, trace round the boundary in one 
direction, say along the path ADC. Join AC. Then, if the 
tracer be taken along CA the value of the area ADC may be 
calculated on the foregoing theory. If now the tracer be brought 
back along AC and thence via B to A, the wheel will indicate 
precisely the same as if the path ADCBA had been taken, for on 
reversing along AC the record on the wheel for the path CA will 
be neutralized. The foregoing theory shows that the sum of 


the readings for the two paths record the number of revolutions 
corresponding to the two areas, and is equal to that for the 
boundary of the figure. Hence, starting at any point and 
tracing the complete boundary gives a record of the number of 
revolutions n, which when added to the zero circle number 
enables the area to be evaluated. 

The Graduation and Calibration of a Tube 

Graduation. The method of graduating a glass capillary tube 
described below is one which could be employed generally in 
the etching of scales on glass. 

A length of glass capulary'tube is coated with a thin layer of 
paraffin wax, by warming the glass and applying a small block 
of wax to the heated surface. The waxed tube is again gently 
heated, in the Bunsen flame, and rotated so that when cooled it 
is evenly coated with a thin wax layer. 

The tube is then clamped to a board at such a height that the 
upper surface is approximately on the same plane as a metal 
scale which is fixed in line with the tube, and at the other end of 
the board. 

A beam compass is arranged so that the two needle points 
are from 50 to 100 cms. apart, depending on the length of the glass 
tube. One needle point is placed in a cm. graduation of the 
scale and the other needle point is drawn across the wax coating 
of the tube, removing a straight line of wax. 

The beam compass is moved 1 millimetre, and a short scratch 
again made on the wax coating. This process is repeated until 
the required length is marked in this way, the whole cm. marks 
and 5 mm. marks being made larger than the rest. 

By means of a steel point the cm. divisions, o, 1, 2, etc., are 
scratched on the wax coating. 

To etch this scale on the glass, a swab of cotton wool, fastened 
at the end of a stick, is dipped into hydrofluoric acid, and then 
applied to the wax-coated tube. Where the scratches of the 
graduations and numbers have removed the wax, the glass is 
etched by the acid. 

In this process care is taken, of course, to avoid any of the 
acid touching the skin or clothing. , 

One end of the tube is also given several scratches, and is 
covered with acid at the same time as the scale. After ten 
minutes or so examine one of these scratches near the end of the 
tube, by scraping away the wax : if the glass is sufficiently 
etched, the whole process is stopped ; if not, leave for a few 
minutes and then examine a second test scratch near the end, 
and so on until the etching is complete. Wash off the acid with 


tap water and remove the wax, the last traces may be removed 
with turpentine or xylol. 

A little rouge, or lamp-black and shellac, rubbed over the 
etched scale brings out the markings. 

Calibration of the Tube. 

The inside of the tube is cleaned either by the usual method, 
using caustic soda, alcohol, ether and tap water, or by immersing 
it for about twelve hours in a solution of potassium bichromate 
and strong sulphuric acid (equal parts). The tube is finally 
washed well with tap water and dried. 

A thread of mercury, about one-third the length of the tube, 
is drawn into the bore, e.g. ip cms. for a tube 30 cms. long. By 
means of a small length of rubber tubing attached to one end of 
the graduated tube, the position of the thread may be varied at 
will by altering the pressure within the tube. 

The mercury is first adjusted so that one end comes near the 
zero graduation. By eye estimation the position of the other 
end may be determined to £ of a subdivision ; hence the length 
of the thread when in this position is obtained. The mercury is 
then moved, by gently blowing through the rubber tube, until it 
occupies the central third of the tube, and its length is again 
estimated in scale divisions. The length taken by the mercury 
is finally measured in a similar manner in the remaining third 
of the tube. The volume of mercury is constant, so therefore 
the length indicated will depend on the average cross-section 
of that part of the tube filled with mercury. Thus this pre- 
liminary test will show the general form of the bore. 

A short thread of mercury, about 1 cm. long, is next introduced 
into the tube, replacing the 10 cm. length. This is moved to 
occupy approximately the space between the o and 1 cm. gradua- 
tion. The tube is arranged horizontally on a sheet of mirror 
glass on the platform of a travelling microscope. The readings 
on the scale engraven on the glass tube, corresponding to the 
two ends of the thread, are taken by means of this microscope. 
For this, the cross-hairs in the eyepiece of the microscope are 
turned so that one is parallel to, and the other at right angles to, 
the length of the tube. The intersection of the cross-hairs is 
brought into coincidence with the meniscus, so that one cross- 
hair appears tangential to it. The vernier reading of the micro- 
scope is noted. It is then moved towards the middle of the 
mercury thread until a graduation on the tube is seen. The 
difference between the vernier reading under these circumstances 
and the last readings gives the distance between the end of the 
thread and the glass tube scale reading. The distance between 
adjacent scale readings on the tube is measured in like manner. 


From these readings the length of the mercury projecting beyond 
an engraved division on the tube may be calculated in terms of 
the graduation of the tube, and the length of the thread of mercury 
may be measured in these units. 

The mercury is moved to occupy the space between the i and 
2 cm. graduations, and again measured. This is repeated along 
the length of the tube, and the result entered as in column 2 of 
the table. The mean value of this length is obtained, and in 
the third column the difference between the observed and the 
mean value is tabulated for each part of the scale (p. 44). 

From these observations we may calculate the correction to 
be applied at each part of the scale to convert the scale readings 
to the corresponding volume readings. Thus, if a tube is divided 
into 20 cms. the mean value of the thread length as calculated 
from column 2 gives the reading for all parts in a uniform tube 
which is 20 cms. long, i.e. the difference between the length of 
the thread between o and I as observed, and the mean value 
gives the correction to be applied to the scale reading to correct 
it to the true volume of thread equal to ^ of the total volume of 
the bore. 

For the bore between 1 and 2 cm. graduations, column 3 again 
gives the correction to be applied to this small part of the bore. 
To correct the total length from o to 2, to give -£$ of the volume, 
the sum of the first two terms in column 3 must be added alge- 
braically. So, to find the correction for any scale reading, the 
sum of the third column must be taken up to and including the 
difference term for that reading. 

This is done with the figures for a tube tested in this manner 
and the correction entered in the last column. 

With the corrections so obtained a correction curve should 
be drawn, plotting the correction as ordinate and the scale- 
reading as abscissae. 

This kind of calibration is often required in thermometry, 
where equal volume expansion of mercury is employed. In 
such a case the scale-readings will not be uniform. In calibrating 
the thermometer the instrument maker fixes a few points along 
the stem by comparison with a standard thermometer. This 
method automatically makes some allowance for the change in 
the cross-section of the tube. 

In the example taken the stem was divided into 1 cm. lengths 
for calibration : it is doubtful whether much is gained by such a 
small subdivision, but is used in the example above to bring out 
the principle involved. In practice, the use to which the tube 
is to be placed determines the number of points tested along the 
tube. With a thermometer 10 points along the length will be 
ample for most purposes, i.e. with a o° to ioo° thermometer, 


the thread detached should be the length of io°C, or about 
3 cms. This is described in greater detail later ("Thermometry "). 






— I 



- -044 

I — 2 


— *004 

- -048 

2— 3 

I '000 

— -004 

- -052 

3— 4 


+ *02I 

- -031 

4— 5 


+ -046 

+ -015 



— -004 

+ -on 



+ -016 

+ -027 


I -000 

— -004 

+ -023 


I -000 

— -004 

+ -019 

9 — io 


— -004 

-f -016 

10 — II 


— -004 

+ -on 

II — 12 


— -024 

- -013 



— -004 

— -017 



+ -016 

— *00I 



— -014 

- -015 



+ -021 

+ -006 

16 — 17 


— -004 

+ -002 



— -004 

— -002 



— -004 

— *oo6 

19 — 20 


+ -006 




The Balance 

The balance, as seen in fig. 20, consists of two pans 
suspended by knife-edge supports, KK, from the ends of equal 
arms of the beam ST, which is pivoted on a central pair of knife- 

The central knife-edge is made of agate and rests on small 
plates of the same material, and KK support such plates. 
The free and sensitive movement of the beam depends upon the 
sharpness of the knife-edges. It is therefore important that 
they should support weight only when in use. To release the 
knife-edges the central pillar Q supports an ' arrestment ' A. 
The fixed arm A carries at each end two points which fit into 
a pair of cups on the upper agate plates at K. When the beam 
is lowered the weight is taken from the knife-edges by this means. 



When releasing the beam, the latter should be in the horizontal 
position, i.e. the beam is only arrested when two outer knife- 
edges are opposite the supporting point of the arrestment. In 
such circumstances all three knife-edges are released with an 
absence of jolting. 


Fig. 20 

In the ordinary way the centre of gravity of the beam, etc., 
is just under the line of support. The lower the centre of gravity 
the less sensitive the instrument. The position of the centre 
of gravity may be varied by an adjustment of the 'gravity 
bob ' B. 


The two masses W may be adjusted to change the rest positions 
of the pointer which moves over the small scale at the base of 
the pillar Q. > 

A fuller account of the balance will be found in " The Theory 
of the Physical Balance," by J. Walker. 

In using the balance, several simple things should be 
remembered, viz., before using, dust the pans with a camel- 
hair or similar soft brush ; arrest the beam before changing 
masses in the pans ; use the rider to start oscillation. Never 
touch the pans or ' weights ' with the fingers, or place chemicals 
or wet vessels on the pans ! Final observations should be 
performed with the case closed. 


The Oscillation Method of finding the Rest Position of the Pointer 

The arrangement of knife-edge supports ensures that the 
friction is reduced to a minimum. Consequently, when the beam 
is oscillating it will have a long period of swing which is very 
slightly damped, i.e. the pointer moving just clear of the scale 
will take a long time in which to come to its final position of rest. 

This rest position may be estimated by the method of oscilla- 
tion as follows : Suppose the scale be graduated from left to 
right into, say, 20 graduations. The end of the pointer, being 
arranged as near to the scale as possible, is viewed by the eye or 
by the aid of a lens, taking care to avoid any slight parallax 
error. The position of the turning point should be estimated 
to at least J of a small division — if possible to T V of a division. 
Seven such turning points should be obtained — three on one side 
and four on the other side of the zero position. The mean value 
of the three on the one side and also the mean of the other four 
should be obtained. Then the mean value of the two means 
gives the position of rest. (See example in the table below.) 

For such damped oscillation the mean of an even number of 
observations on each side of the zero would give a value biased 
towards the side of the zero of the first observed turning point ; 
so an odd number is taken as stated. For if we assume that 
the damping is small and that the first swing through the rest 
position is to the left, we have, from a consideration of such a 
lightly damped oscillation, the angular deflection, 0£, the first 


swmg to the left is after a time - and is (see page 28), 

8l = e- A * 
where A is a constant, O the undamped oscillation. 

This is, under the conditions of very small damping, 
01 = O (i - A ~) to the left. 
The deflections are therefore : 




, (x_aT) 


eo (x-A ? ) 


mean O (i - A^) O (* - A^ 

i.e. the mean is o since the mean position left and right is an equal 
distance to each side of the zero. 

Zero Position of the Unloaded Balance 

As an example of the method of oscillation, the following 
determination of the zero position for the unloaded balance is 

The beam is released from the arrestment and given a slight 
oscillation over, say, about 5 divisions of the scale on each side 
of the zero. After a few swings the oscillation should be steady, 
and seven readings, are taken, as under : 



I. 8-8 
3. 9-2 

5- 9'5 

7. 9-9 

2. 12-0 
4. II'l6 

6. n-2 

Mean of left hand reading = 9-35 

Mean of right hand reading = n-6 

~ . . t , Q'^5 + H'6 

Zero positions unloaded = ^-^ = 10-475 

A mean value of three such determinations gives the rest point. 
In general, call this rest position for no load X. 


Sensitivity of the Balance 

The sensitivity of the balance is denned as " the angle through 
which the beam will turn for one milligramme difference in load in 
the two pans." 

It is usual in practice to measure the sensitivity in terms of 
the movement of the potater over the scale. 

This should be measured for no load, by adding one milligramme, 
by means of the rider, to one pan and noting the position of rest 
by the oscillation method as before. 

Load each pan with increasing equal masses up to the limits 
specified for the balance, and for each mass find the sensitivity 
i.e. change of rest position for one milligramme on one pan. 

If the beam were rigid and the knife-edges truly in the same 
line the sensitivity-load curve would be a straight line parallel 
to the load axis. 

For increasing loads there may be, however, a slight depression 
of the knife-edges at the pan supports, and a change in sensitivity 
as a result. 

Find the sensitivity-load curve for the balance and note the 
region, if any, of maximum sensitivity. The value of the load 
at which there is maximum sensitivity depends upon the use for 
which the balance is designed. 

Method of Gauss or Double Weighing 

As an example of this method we will consider a determination 
of the number of grammes equivalent to one ounce troy ; the pro- 
cess can naturally be repeated with any other unknown mass. 
Place the ounce troy in the left-hand pan and add ' weights ' 
(grammes) to the right-hand pan until the pointer remains on the 
scale when the beam is released ; the rest position of the oscillat- 
ing pointer is estimated in the manner previously used. Let 
this be y on the scale. y 

Now find the sensitivity of the balance at this load, by adding 
one milligramme and proceeding as before. Suppose the sensibility 
for this load is s. 

The mass of the ounce troy in grammes is the mass (to the nearest 

centigramme) in the scale pan + ( ^7*) m^S 1 "* 11111165 = M i» 
say, when x is the zero reading for no load. 

The ounce troy is then transferred to the right-hand pan and 
the process repeated by adding ' weights ' to the left until, to 
the nearest milligramme, M 2 , the mass as so compared is obtained. 

Suppose that the two arms of the balance are of slightly different 
length, the left-hand arm being a cms. and the right b. 


Then W being the true mass of the ounce troy, neglecting 
buoyancy, > 

aW = M x 6 
«M 2 = Wft 

Hence W = VMA 

The difference between M x M 2 and W will be very small, so that 
we may take as an approximation 

W = i(M, + M,). 
Borda's Method, or the Substitution Method 

A second method, quite as accurate as the double weighing, 
is a simple method of substitution. It eliminates equally well 
the errors due to unequal length of the arms, etc. 

The ounce troy (the unknown mass) is placed on a scale pan 
and lead shot is used to counterpoise it. The position of rest 
when the counterpoise is complete is noted by the method of 

The ounce troy is now removed and replaced by standard 
masses until balance is again obtained. From the sensitivity 
of the balance for this load we may estimate very readily to a 
milligramme the mass which has exactly substituted the ounce 
troy. Thus the mass is obtained, avoiding errors due to faulty 
construction of the balance, etc. 

Buoyancy Correction. 

When discussing methods of weighing'no account was taken of 
the buoyancy of the air on the ' weights ' and the mass to be 
compared with them. 

Suppose, as before, we find that W is the mass of the body, by 
one of the methods above ; the true mass allowing for buoyancy 
we will denote by M. 

Suppose that in the determination of W we used copies of 
standard masses made of a substance of density D. 

Let p be the density of the ' unknown ' mass and o the density 
of the air. 

We have really compared (1) (the true mass of the body M — 
the buoyancy on the mass) with (2) (the mass W of the ' weight ' — 
the buoyancy on the ' weights '). These two quantities are equal, 


i.e. M =w(i -~\ 


-M' + -;-»--) 

neglecting -= cf . i. 

= W + W ( I p-D> 

The observed value of W has therefore to be corrected by the 
factor W g -i).. 

This correction depends on the density of the ' weights ' and 
the substance and the density of the air. 

For most purposes the density of the air may be taken as •0012 
grm./ and for general use a table may be calculated giving 

the value of the correcting factor I =- J o for the two common 

materials used in the manufacture of ' weights,' brass and 
aluminium, taking the density of brass = 8-4 grms./ 
Aluminium = 2-65 grms./ Thus : 






' WEIGHTS ' D =2-65 

V p 2-65/ 







Kinetie Energy of a Body Rotating about an Axis 

Let ABC (fig. 21) be a section of a body by a plane at right angles 
to the axis about which it is rotating, O being the point of inter- 
section of this plane and the axis. 

,FlG. 21 

If we imagine the body to be subdivided into a large number 
of very small particles of mass m x , m 2 , m 3 , etc., distant r v r 2 , r 3 
cms. from the axis, it is evident that when the body rotates 
each particle will move with a velocity which depends on the 
distance r from the axis. 

Consider one such particle at P, of mass m and distant r cms. 
from O. If the body rotate with a uniform angular velocity w, 
in the direction of the arrow, and if v is the velocity of P in the 
path, we have 


- = w. 

The kinetic energy of this particle is %mv z = \mw % r i . For 

all such particles the total kinetic 'energy of the body is therefore 

^MjW^r 2 + ^m z w 2 r 2 z + J^gze'Va 2 + . . . . 

i.e. Kinetic Energy = H\w 2 mr* 

= \w* Emr* 

The sum of such quantities as mr 2 , taking every particle 

throughout the body, is denned as the moment of inertia of the 



body about the axis through O. If we denote this by I , then the 
kinetic energy of the body is 

ll wK.... (i) 

thus I replaces the mass, and tfre angular velocity replaces 
linear velocity in the corresponding case for linear motion 
where K.E. = %mv 2 . 

In the same way if the axis passed through the centre of 
gravity ; I being the moment of inertia about an axis through 
the centre of gravity, K.E. = $Iw 2 , 1 being of a different magni- 
tude from I . 

To express the moment of inertia of a body about an axis in 
terms of the moment of inertia about a parallel axis passing through 
tfee centre of gravity, we proceed in the following manner : 

In fig. 22 let ABC be a section of the body at right angles to 
either axis. Let G be a section of the axis passing through the 
centre of gravity and O be corresponding point of intersection 
for a parallel axis, the distance OG being fixed and equal to 
a cms. 

Fig. 22 

Consider at any point P a small particle of mass m gm., OP 
being r cms. 

The contribution of this particle to I is mr 2 . Now produce 
OG to D, and from P drop a line perpendicular to OD meeting 
it at D. 

f 2 = OP* = OD 2 -f- PD 2 = PD 2 + DG 2 + GO 2 + 2OG • GD 

= PG 2 + a 2 + 2a • GD 
Thus I = Em (PG 2 + a 2 + 2a • GD) 

= SwPG 2 + Swa 2 + X2ma • GD. 
Now SwPG 2 = I, the moment of inertia about a parallel axis 
through the centre of gravity. 

Sma 2 = a 2 Zm = a 2 M, 


where M is the total mass of the body. Further, the expression 
2aSGD«w = o, for by the definition of the centre of gravity the 
sum of the moments {GD-m) throughout the body, about an 
axis through G, is zero. 
Thus , . I = I+^ 2 m (2) 

Radius of Gyration. 

We have defined the moment of inertia of a body about an 
axis as 2#w 2 . Now if the whjole of the mass of the body were 
concentrated at one point distant K from the axis, we should have 

I - K 2 M. 

If the distance K were s o chosen that 

K 2 M = Emr 2 = 1 v - 

it is called the ' Radius of Gyration ' of the body about the 
axis of rotation taken. 

As with the moment of inertia K has different values depending 
upon the axis chosen. 

Moment of External Forces. 

Considering a rotating body as before, we may readily deduce 
an expression for the moment of the external forces applied to 
the body to impart a definite angular acceleration. 

Imagine a force applied to such a body as shown in fig. 21 

Suppose the body to be subdivided into small particles as before, 

of which one at P has mass m and is r cms. from O. Then 

v, the velocity of the particles, is given by 

dd . 
v = r • -r. or rO 

at * 

the acceleration of the particle in its path is 

dv d*B ■■ 

dt= r aW ==rd 

this is occasioned by a force mr§, whose moment about O is 

mr 2 0. For the whole body to rotate with this angular acceleration 

the total external couple applied is thus 

Sw 2 = QXmr 2 = 10 (3) 

Thus the moment of external forces applied to the body is I'd. 

Calculation of the Moment of Inertia for a Solid about any Axis. 

The numerical value of the moment of inertia of a solid about 
any axis may be readily obtained by integration. Having 
calculated this value for an axis passing through the centre 
of gravity, the corresponding value for the case of the body 
suspended through a parallel axis may be obtained by adding 
the term Ma 2 as shown on page 52. 


We will consider an example of such calculations which is 
often employed, especially in magnetism, and which illustrates 
the points already considered. 

Calculation of the Moment of Inertia of a Rectangular Rod about 
an Axis at Right Angles to its Length and passing through the Centre 
of Gravity. 

Let ABCD be the rectangular bar, with centre of gravity at G, 
and supported by an axis KK 1 passing through G, normally 
to the face AD, fig. 23. Let 

M be the mass of bar (assumed to be uniform) 
p the density of the material of the bar 
ol the length of the bar 
2& the breadth of the bar 
2d the depth of the bar. 

-y b 

Fig. 23 

Let G be the origin of a system of co-ordinate, the z axis coincid- 
ing with the axis of rotation : the x and y axes being at right 
angles to this, the x axis being parallel to the length, the y 
parallel to the breadth. 

The most convenient method of finding I the moment of 
inertia about KK 1 is to consider firstly a very thin section of 
the bar cut at right angles to the x axis, and of thickness 3x. 
Through the centre of mass G 1 of this section, imagine an axis, 
LL 1 , parallel to KK 1 . TJie moment of inertia of this section 
about KK 1 is equal to the moment of inertia about LL 1 plus the 
product of the mass of the section and x 2 , where x is the 
distance between G and G 1 . Imagine a very small rectangular 
portion PQRS of EFHI, y cms. from the axis and of width dy 
(fig. 24). 


The mass of this parallelepiped is [dx dy 2cC\ p. The moment of 
inertia of the section EFHI about LL 1 is therefore 

fzddxdy gy* = 4^- p • dx. 
J -h 3 

About KK 1 the moment of inertia is therefore 

***'"** + 2d2b 9 xHx. 
Hence the total moment of inertia of the whole body about the 

axis KK 1 is * 



pdx + 4 bd x 
(6 2 + I 2 ) 

'■pdx J 

- j(b 2 + 1% 


When the moment of inertia of a body cannot be conveniently 
calculated it may be found experimentally by imparting to it 
a known amount of energy and observing the resulting rotation, 
or, if the body is small, the method of the moment of inertia 
tableonay be employed, whereby the change in moment of inertia 
in a given system, due to the body, may be directly calculated. 

The following typical" experiments will make these methods 

Moment of Inertia of a Fly-wheel 

To find experimentally the moment of inertia of a fly-wheel 
about the fixed axis of rotation a mass is attached to the axle 


of the fly-wheel by a cord which is wrapped several times round 
the axle. When the mass descends, it causes a rotation of the 
wheel. The mass in its descent loses a definite amount of 
potential energy. Neglecting friction for the moment, this loss 
is equated to the gain of kinetic energy of the mass and the fly- 
wheel, and an equation results in which all the terms are known 
or measurable except I, the moment of inertia of the wheel and 
axle about the fixed axis. 

Fig. 25 

The wheel might be supported on a horizontal or vertical 
axis. The two usual types met with are seen in fig. 25. The 
process to find I is the same, so we will consider one of them — 
the vertical axis type. 

The mass m is attached to the axle at a point where there is 
either a hole or a pin. If there is a small hole in the axle as 
at P, then to the end of the cord a small ' pin ' is attached. This 
can be made from a short length of suitable-sized brass wire 
to fit easily in the hole, or if the axle has a pin projecting, a loop 
is made at one end of the string. The length of the string is so 
adjusted that when m is on the floor, or whatever solid object 
is to arrest it in its descent, the other end of the string may be 
just attached to the axle. So that when the mass descends, 
the moment it is arrested, the string leaves the axle. 

If w be the angular velocity imparted to the wheel, and r 
be the radius of the axle, the velocity of the mass m just 
before striking the floor is rw. So that, neglecting friction, we 

have mgh = fynrV -f £Ize> 2 (4) 

where h is the distance through which m has fallen. 

To measure the angular velocity, a chalk-mark is made on the 
circumference of the wheel, in a position which can be seen 
the moment the mass touches the floor. The number of revolu- 
tions, n, made by the wheel after the mass becomes detached 
is counted by observing the chalk-mark. The time taken for 
the wheel to come to rest whilst completing the n revolutions is 
also observed (t sees.). 


The wheel is finally brought to rest by the frictional forces 
acting against it. If this frictional force is constant, the wheel 
is uniformly retarded. It commences with a definite angular 
velocity and finishes with zero angular velocity, so that the initial 
velocity is double the average velocity. 

Now the average angular velocity = —7- radians per second, 

i.e. w —- L -r* 

The linear velocity of the mass m is - — 

Another method, though inferior to the above, is to observe v 
directly by timing the descent of the mass. If the mass descends 

the distance h in t 1 sees, the average velocity is — ' and the final 

velocity is -pr- 

As we have noticed above, the frictional forces are not always 
negligible, so that, for a more accurate determination of I, 
allowance must be made for the energy lost in overcoming friction. 

Let there be n x revolutions of the wheel during the descent of 
the mass, and let / ergs be the energy per revolution used in 
overcoming the frictional forces, then the total energy expended 
in this way is n-J. 

mgh = fwu 2 + \lw 2 + nj. 

Now we already know that the energy possessed by the rotating 
wheel, \Iw z , is used up in overcdming friction in n revolutions, 
i.e. fn = %lw 2 


f = ^Iw\ 


i.e. mgh = \mrhiQ*+ |I*Wi +— M (5) 

Experimental Details. 

Arrange the cord round the axle so that throughout the whole 
of the unwinding the cord from the axle to the pulley, T, is 
practically horizontal, or at right angles to the axle. 

Bring the mass m so that the bottom of it is level with a fixed 
point, and the string of such a length that it fulfils the conditions 
already stated. 

The distance h from the fixed point to the floor is directly 

* measured. The number of revolutions the wheel makes whilst 

the mass is descending may be determined by making a chalk- 


mark on the axle and allowing the mass to descend slowly, 
counting the number of revolutions (%) during the descent. 

The mass is once more wound up and allowed to fall freely. 
When it is heard to strike the floor a stop-clock is started and 
the number of revolutions of the wheel before being brought 
finally to rest is counted, i.e. n and t are observed : m the 

mass is known and w = ^-7—-, hence the value of I is calculated 

by the aid of equation (5). 
The experiment is repeated two or three times with the same 

mass and the mean value of — taken. 

I is further checked by repeating with two other masses m 1 
and m 2 . 

The cord used should be of small diameter compared with the 
diameter of the axle, otherwise the value of r in equation (5) is 
the sum of the radii of the axle and cord. 

Rolling Bodies 

The two following experimental methods of finding the moment 
of inertia of a body about a given axis depend upon observations 
of rolling bodies. 

The energy of a rolling body may be very simply obtained. 
Consider, for example, a cylinder rolling with a uniform linear 
velocity v cms. per second (fig. 26). 

AB, the line of contact of the cylinder and the plane on which 
it rolls may be regarded as the momentary axis of rotation of 
the cylinder. 

The kinetic energy of the body is therefore given by 

£ze> 2 (moment of inertia of the cylinder about the axis AB) 
where w is the angular velocity of rotation. 

Now, if a is the radius of the cylinder, m its mass, and I the 
moment of inertia about a parallel axis through the centre of 
gravity; the kinetic energy is : 

\ w 2 (I + ma % ) 
= |lrc; 2 + \mv* (6) 


That is, the kinetic, energy is equal to the sum of the kinetic 
energy of rotation and translation. 

Wheel and Axle on an Inclined Plane 

The moment of inertia of a wheel and axle about an axis 
passing through the centre of gravity and parallel to the axle 
may be obtained by observing its descent down an inclined plane, 
and applying equation (6). 




/// / / 



Fig. 27 

The method of rotation will be apparent from a consideration 
of the section, fig. 27. R and R^are rails supported on a hollowed 
inclined plane. The axle of the* wheel rests upon the rails. The 
whole plane may be inclined at any angle to the horizontal. 
For each inclination the wheel and axle is allowed to roll down a 
measured length, I cms., of the plane, in a time which is measured 
by means of a stop-clock (t sees.). 


Fig. 28 

If the vertical distance between the starting position and the 
finishing position be h cms., fig. 28, and v be the final velocity 
acquired in the descent, we have, equating potential energy lost 
to kinetic energy gained, 

mgh = %mv 2 + £1 M (7) 

where m is the mass of the wheel and axle, and a the radius of 
the axle. 

The body starts from rest and moves with a constant accelera- 
tion ; the final velocity is therefore twice the average velocity. 

This latter is equal to -, i.e. v = —• 

The plane is adjusted by suitable means to one fixqd inclination. 
The wheel and axle is placed at a convenient marked starting 


point on the rails. The position of the centre of the axle is noted 
by means of a vertical-reading simple cathetometer. The position 
of the centre of the axle is also noted when the wheel is against 
the stop at the other end of the plane. The length I along the 
plane between these two points is measured directly. The value 
of the mean time of descent for three experiments is obtained. 
The mass m is also obtained by means of a spring balance or an 
ordinary balance which is capable of weighing such a mass. 
a is obtained in the usual way by means of vernier callipers. 


Hence, from (7), substituting the value - for v, 



The experiment is repeated for several values of h and the 
mean I is obtained. 




1 w 



Fig. 29 

Moment of Inertia 0! a Disc Supported on Strings 

A disc, usually made of wood, is suspended by means of a metal 
axle on two strings, as shown in fig. 29. The string is wound 
evenly on the axle AB on both sides until as much string as 
possible is wound up. If now the axle and disc, of mass m 
grammes,' is released, it will descend until the whole of the cord 
is unwound ; it will then rise again due to the string being wound 
on the axle in the other direction. 

Suppose that from the starting point to the lowest point 
reached the distance the centre of the axle moves is h cms., 
and that the linear velocity at the moment when all the string 
is just unwound is v cms. per sec, then, if r is the radius of the 
axle, we have as the energy equation (equating potential energy 
lost to kinetic energy gained) : 

mgh = %Iw* + \mv*, 


I being the moment of inertia about an axis passing through the 
centre of gravity and parallel to the axle. 

w the angular velocity at the lowest point ^s = -, 

i.e. mgh —-S-j- + fmv 2 . 

I = (mgh - frnv 2 )^- 

If the time of descent of the disc is t seconds, the average 

velocity is - cms. per sec. 

v, the final velocity is therefore — . 

i — '*(£•- *) 

Experimental Details. 

Weigh the disc, then measure the distance between the position 
of the centre of the axle in the starting position and the final 
lower position. 

The value of r is equal to the sum of the radii of the axle and 
the cord which supports it, unless the cord has small radius 
compared with the radius of the axle. These radii are measured 
by means of a micrometer screw. 

The cord is wound evenly on the axle until the disc is at the 
starting point. Care is taken to ensure that the axle is horizontal, 
otherwise the disc fouls the cords in descent. 

The time of descent is measured several times by means of a 
stop-clock, and the mean value taken. 

The Bifilar Suspension 

In order to determine the moment of inertia of a body about 
an axis passing through its centre of gravity, we may make use 
of a bifilar suspension of known or measurable dimensions. 
The body is suspended with the axis of rotation vertical, and the 
time of vibration of the system, T, obtained by observing the 
time of 40 or 50 complete swings. 

If, for example, we wish to determine the moment of inertia 
of a cylinder about an axis through the centre of gravity, the 
cylinder is supported in two wire stirrups CE and DF (fig. 30), 
which hang at the ends of two very thin wires AC and BD, which 
are fixed at A and B, AB being 2d cms. The distance between 
C and D remains fixed and equal to 2d 1 cms. 

When the body is displaced slightly in the horizontal plane 


it will perform oscillations whose periodic time T may be ascer- 
tained as already shown. 

If m is the mass of the cylinder and / is the length of the wire 
AC or BD, we may readily see that 

t = 2x /te: 

^ mgdd 1 
For, let the tension in the string be t dynes and O be the mid- 
point between A and B. 

j — i 


Fig. 30 

When viewed from above, fig. 31 (a) represents the relative 
positions of the four ends of the wires. When displaced, the state 
of affairs is seen in fig. 31 (b). Where A 1 and B 1 are projections 
of A,B, fig. 30, on the horizontal plane through CD, and C 1 , D 1 
are the displaced positions of C, D. 

Consider the forces at D. Due to the tension on the string 

there is a force t, which has a horizontal value t cos a, where a is 

the angle between the string BD 1 and the horizontal. Now 

B^ 1 
cos a = — - — » for, in reality, the point of suspension, B, is above 

B 1 , BB^ 1 being a right-angled triangle. 
So that along D^ 1 in the horizontal plane there is a force 

/-«^ (8) 

Of this force the component at right angles to the displaced 
body, i.e. C^ 1 , is effective in restoring the body to its original 
position, the component along the direction OC having no 
turning moment. 

From B 1 draw B*E at right angles to C 1 D 1 . The component 


EB 1 
normal to OD 1 is / sin EDW =%i^r = / 1 ' sa y» 

from (8) f 1 = EB 1 '- (9) 

A similar force acts at C 1 , constituting a restoring couple of 
moment/ 1 - OD 1 . 
The restoring couple is, substituting value of f 1 from (9), 

EB 1 * 

OD 1 . 

A O. 


D (a) 


Fig. 31 

If the small angle of displacement, D 1 OB 1 = 0. 

EB 1 = OB sin = d sin 0. 

OD 1 = 2d 1 . 

2dd x 
/.Moment of the couple = —j- t sin 0. 

is usually made very small, so we have for the value of the 
restoring couple: 

2dd 1 mg ^ 


= **%%. e (10) 


for when d is not very different from d 1 1 =£= — . 

We have already seen (p. 53) that the moment of external 

d 2 6 
forces acting on a suspended system is I -^ ; this quantity is 

equal and opposite to the restoring couple, 


T d*d _ mgdd 1 

dt* I 

This will be recognized as the equation of simple harmonic 

motion (p. 25), as we have the angular acceleration = It— X 

the angular displacement, ° . being a constant. 

The time of vibration is therefore 

27C / II f TJ \ 

It is interesting to note that for bodies of the same dimensions 
and of uniform density the value of T is the same. For, 
let p be the uniform density, then I = Ewr 2 . Consider the 
body divided into small volumes v, : Smr a = Xvpr 2 = pE vr % . 

Also m = Spv = pSv 

—tt£ ♦ » i.e. independent of the density. 

The above experiment may be carried out using a metal and 
a wooden cylinder as the supported body. 

If the dimensions of the wood and metal cylinder are practically 
identical, the time T will be found to be the same within small 
limits. Suspend each in turn and find T by timing 50 swings, 
or by the method of page 118, and, from the formula above, 
calculate I. 

For such a regular solid an independent calculation gives a 
second value of I, which should agree very nearly with that 
already obtained. 

/ja *2\ 

For a cylinder, about the axis taken, I = M( 1 — J where 

2L is the length of the cylinder, r the radius. 

The formula may be further tested by varying d, d 1 and I. It 

will be found that T 2 is proportional to t-^. 

Moment of Inertia Table 

Fig. 32 shows the essential features of the moment of inertia 
table. AB is the table suspended by a fairly stout wire, K, from 
the overhead frame. The circular table supports three or more 
masses which just fit into a groove, concentric with the circum- 
ference. When the wire support is vertically above the centre 
of gravity the masses, W, may be moved round the groove into 
any position without altering the moment of inertia of the whole. 
The masses are arranged so that the table lies horizontally. In 
that case the axis is through the centre of gravity and is normal 
to the surface of the table. If now the table is given a slight 
twist, a restoring couple is called into play in the wire, equal 
to, say, t per unit angular displacement, and the result is 
oscillations about the axis of support, of periodic time 

T = 2 ,yx 

^ T 

where I is the moment of inertia of the system about the axis 


of rotation ; for) suppose the table be twisted through an angle 0, 
the restoring couple due to torsion is t0. We have already seen, 
page 53> that the moment of external forces on such a rotating 

body is 10" where 9 is— 57a • This couple is opposed by t0 : t0 and 

10 are equal and opposite ; i.e. 

I 9 = - T0 








This was shown in the introductory chapter (p. 25) to repre- 
sent simple harmonic motion whose periodic time, T is given by 

T= /-=V- 


If now a regular shaped body is placed symmetrically on the 
table at C, so that its centre of gravity is vertically above the 
centre of gravity of the table and therefore in the previous axis 
of rotation, the time of oscillation for the loaded table is 

\= 2rt y 


where k is the moment of inertia of the regular body about the 
axis of oscillation : k may be calculated directly from the mass 
and dimensions of the body. 



We have, therefore, from the two equations above : 
8 _ 4" 2 (I + k) . Ts _ 4 * 2 (I) 

„ T, a I + k . k 

Hence -^ = —~- = i + j> 

or I = k T 2 _ Ta (12) 

If, then, a bcdy of unknown moment of inertia about a given 
axis is placed centrally on the table with the centre of gravity 
in the axis of oscillation, we may find the value of I lt its moment 
of inertia abbut this vertical axis passing through the centre of 
gravity, by timing the oscillations of the table when loaded by 
the body. If T 2 is the time of complete swing, 

Ts _ „ yr±n, 

and we have 

t =2«y-» 

i.e. from these two equations : 

T a T2 

i. = il -r^' 1 (I3) 

Substituting, from equation (12) above, 

T 2 T* T2 T 2 T2 

x i Y 2 x a — T 2 ~ T 2 — T 2 ^™' 

Experimental Details. 

The time T for a complete swing of the table is obtained by 
timing as many swings as possible, or by the method of p. 118. 

To find the value of I, the moment of inertia of the unloaded 
table about the wire as axis, a regular solid, such as a plain 
cylindrical ' weight ' is employed. The mass should be fairly 
heavy so as to cause as big an alteration in T as possible. A 
two-kilogramme ' weight ' is of the order to employ with the 
apparatus described. 

To ensure that this standardizing mass is arranged with its 
centre of gravity over that of the table, the lead weights W should 
be adjusted so that the table swings horizontally. Any alteration 
in the position of the masses, W, will not alter I, so long as the 
table is horizontal and the axis of oscillation is vertically through 
the centre of gravity. 

Having obtained T and T v using a cylindrical regular ' weight/ 


the value of k should be calculated. For such a flat cylindrical 
object, k = — - where a is the radius and M the mass of the 

' weight.' 
By equation (12) I is calculated and the table standardized. 
Other regular solids may now be used and I, obtained by 
equations (13) or (14), and then by calculation a check is obtained 
on I x as previously obtained. 
The following results were obtained in the above manner: 
Table unloaded Table loaded with Table loaded with 
T 2000 grammes unknown body 

Time 9-375 sees. 97 sees. i5'5 secs - 

radius of 2000 gm. ' weight ' = 675 cms. 
k _ 2000 x 675' =45 . 5e2gm , cm *, 


Ii = 45*502 X 

i5*5 2 - 9'375 s 

97 2 _ 9 . 375 2 

Ij = i'ii x io 6 gm. cm 2 . 
I, by approximate calculation, assuming a regular shape to 
the body = 9-86 x 10 5 

The Compound Pendulum. 

Let ABC, fig. 33, be a section of a body, passing through the 
centre of gravity, and at right angles to an axis about which 
it may turn, the point O being the intersection of the axis with 
this plane section. 

The body is at rest when the centre of gravity, G, is vertically 
under O. 


If the body is given a small displacement so that GO makes a 
small angle with the vertical, then, m being the mass of the 
solid, the restoring force has a moment mg OG sin = mm sin 6 
(putting OG = a.) 

We saw (p. 53) that in such a case the moment of the forces 

is equal to Io^. 

*2 2 
i.e. I -=- = — mga sin = — mgad for small angular 

This represents a simple harmonic motion (p. 25), whose 

periodic time, T = 2w \J — — • 

I = I + ma 2 , where I is the moment of inertia about a 
parallel axis through centre of gravity and is equal to mk 2 where 
k is the radius of gyration about this axis. 

Thus T = 2* Jft> + a2fH = 2nJ« a (l 5 ) 

\ mag > g v 3/ 

This result is similar to that obtained for a simple pendulum : 

k 2 
in fact, a simple pendulum of length I = — + a would have 

the same periodic time, T. Such a simple pendulum is called the 

' Equivalent Simple Pendulum.' 

In the case taken, if all the mass of the body were concentrated 

k 2 
at a point, P, along OG produced such that OP = \- a, we 

should have a simple pendulum with the same periodic time. 

The point, P, is called the ' Centre of Oscillation,' O being 
called the ' Centre of Suspension.' 

Now since 

I = \- a 


or a 2 — al + k 2 = o, 

the length a is not the only value for OG, which has / as 

the equivalent simple pendulum, for the above equation has two 

roots, a x and cc 2 , such that, 

<«1 + «2 = l \ (l6) 

ajOCa = k 2 ) 

Since a is one value, a x say, 

k 2 
we have a + oc a = I or oc 2 = I — a = — • 

Thus if the body were supported on a parallel axis through the 


former centre of oscillation; P, it would oscillate with the same 

time T as when supported at O. 

From what has been seen above it is evident that there are an 

k 2 
infinite number of points distant, a and — from G, for any 

k 2 
point on a circle drawn from G as centre and radius a or — 

will satisfy the condition given ; so that any axis parallel to the 
normal at G on the curved surface of two cylinders, of which 
the dotted circles are sections, will be axes of suspension which 
give the same time. 

If the body were supported by an axis through G, the time of 
oscillation would be infinite. From any other axis in the body 
the time is 



a 2 _1_ £2 . 

This has a minimum value when is minimum. 


a 2 +k 2 a* — 2ak + k 2 + 2ak {a - k) 2 + 2ak 

Now = ■= 

a a a 

This is a minimum when a = k. 

The corresponding minimum Tj is T t = 2*:^-— (17) 


and will occur for a series of axes parallel to that through G, 
and on the surface of a cylinder whose axis is the axis through 
the centre of gravity and radius, k. 

An experiment which brings out these facts may be performed 
by using as the body a rectangular rod of brass about 1 metre 
long. This may be suspended on a knife-edge at various points 
along its length. To facilitate such suspension it is convenient 
to have a series of holes drilled along the bar at about 2 cms. 
intervals (fig. 34), 

Level the knife-edge, and suspend the bat at, say, every other 
hole in turn, and time 50 swings at each hole, which is a measured 
distance from the centre of gravity of the bar (which may be 
obtained by simple balancing). Or the holes may be measured 
from one end of the bar. 

Having obtained a set of values for T, and the distance from 
the centre of gravity, plot a curve with the periodic times as 
ordinates and the distances as abscissae. A curve such as shown 
in fig. 35 will be obtained. 

The values of T near the minimum points, MM 1 , should be 
further investigated by taking the time for vibrations in every 

•See also page 118 for a method of timing. 


hole, three each side of the approximate position, and the graph 

Let the line CG, fig. 35, be drawn from C, which represents 
the centre of gravity oX the bar. 

Draw any line EABFD parallel to the axis. This cuts the 
curve in four points, which have the same periodic time, T = BC. 
It will be found that the lengths, FB, BA, and BE, BD, are equal ; 
i.e. FB and BA correspond to radii GQ, GP, and BE and BD 
to radii GO and GR, in fig. 33. 

Take either set in pairs, say BA, BD, these are corresponding 
lengths to a x and <x 2 in equation (16), i.e. AB + BD = /, the length 

= 2 *Vi 

Fig. 34 

of the equivalent simple pendulum. Its periodic time T is 
numerically equal to BC. 

Hence from equation, T 

all factors except g are known, whence g may be calculated. 

If now a tangent is drawn to the curve, such as line LMM^L 1 , 
HM = HM 1 = radius of gyration about an axis through the 
centre of gravity : this may be measured directly. 

Further, by equation (16), k = V^ol^ = VAB • BD, so 
a second value of k may be found. The corresponding periodic 
time is, numerically, the length of HC = T lf say. 

Hence, once more, by equation (1 7), 

V g 
g may be evaluated or, as the direct exact measurement of 
k (MH or M^H) is difficult, as there is some doubt in the general 
case as to the exact location of M and M 1 , this formula be used 
to calculate on third value of k, 

Ti 2 


ft — 75 — 2 ■ 


The mean value of k may be taken and the moment of inertia 
about a parallel axis through the centre of gravity calculated, for 

I = km, 

where M is obtained by direct weighing. 

L- - 

- C 

Ois^ortce. fromC.G D'wsfaoce Tfom OG 
Fig. 35 

Hater's Pendulum 

From the preceding experiment it is obvious that if it were 
possible to obtain, for a rigid body; two parallel axes of suspension, 
along any line through the centre of gravity and on opposite 
sides of it, which have exactly the same time of swing, 
then the value of g could be very well determined by 
measuring the distance between such axes. This distance 
would be equal to the length of the equivalent simple pendulum, 
/. If the equal periodic time about these axes were T, 


= *■<$• 

I and T being measured directly. 

The Kater pendulum is one by means of which this may be 
realized in practice to a very close approximation. 

It consists of a long rod which is provided with two fixed 
knife-edge supports, K and K, and terminates at each end in 
a ' bob,' B and B. Usually, the one bob, B, is made of brass 
and the other of wood. 

M and m are two adjustable masses which may be fixed in any 


position between the knife-edges. Their adjustment serves to 
move the centre of gravity to such a position that the time of 
swing is approximately the same from either knife-edge. 

The pendulum is supported on knife-edge, K and K, in turn, 
and the approximate periodic time, T and T lt is obtained by 
counting swings, timed by means of a stop-clock. 

The large mass M is moved until these times are approxi- 
mately the same. The small mass m serves as a fine adjustment 
to this purpose. 

Having adjusted the masses so that the time for a complete 
vibration is very nearly the same from both knife-edges, it will 
be realized that to obtain exact agreement for T and T x would 
be a most tedious experiment. 

Fig. 36 

However, we can see in the following way that such exact 
agreement is not essential. 

Let a and a x be the distance from the centre of gravity of the 
pendulum to K and K. 

> ag y a 

+ k 2 

ag > a x g 

T*ag == 4*2 (a* + k*) T x 2 a lg = 4* 2 (a^ + k*) 

(T z a-T 1 *a 1 )g=4n*(a*-a 1 *) 

£! = a T2 ~ <*i T i a _ I/H±Il* j. HjzIA 1 a\ 

g a*- ai * ~2\a + a 1 + a - a x ) '-"P*) 

a and a x may be made to differ by a fairly large amount by 
suitable adjustment of the masses, M and w. With a little care 

*p2 np 2 

T and T x may be very nearly equated, and so the term — 

a ■— a x 

becomes small. T and T x may be measured very accurately by 
the method of coincidences, (a -f a-^ may be measured directly 
as the distance between the knife-edges, by a comparison with a 
metal metre scale by means of a comparator (p. 32). The 
important first term in equation (18), (R.H.S.) is thus carefully 


The second term is small and no serious error is involved if 
a and a x are measured from the knife-edges to a point at which 
the pendulum may be balanced horizontally on a knife-edge. 

By this method a very reliable value of g may be obtained. 

Method of Coincidences. 

This method of timing a pendulum consists in hanging it 
by a knife-edge from a rigid support, in front of, say, a seconds 
pendulum of a standard clock, the height of the support so 
arranged that the tails of both pendulums are on the same level. 
At rest, viewed by a telescope from in front, the Kater coincides 
with the seconds pendulum. 

If both pendulums are of the same period and start oscillating 
together, when viewed through the telescope, they appear to 
move as one. If the periods are not the same, they will be 
seen to get ' out of step,' and at one point both will pass a fixed 
reference point together and going in the same direction. This 
will not again occur until one pendulum has gained or lost a 
whole swing. 

Suppose the seconds pendulum makes n complete vibrations, 
each of period T (2 sees.), and the experimental pendulum makes 
(n + 1) complete swings, of period T lt 
Then Tn=T 1 (n + i) (19) 

Tl _ n 1 _ 1 _i_ , v 

T — n + 1 ~" i — n n 2 * ' 

+ n 

Suppose n = 500. 

T *-i-^ + 

T 500 250000 

further terms are negligible. 

Hence T j is obtained in terms of T ; which in the case taken 

is 2 sees. 

Similarly the time about the other axis may be checked. 

In the coincidence method it may be observed that one is 
never quite sure within a few (say m) passages of the pendulum 
which is the correct coincidence. We can easily see that the 
error introduced by this cause is not appreciable when n is fairly 
large. Thus we know that in equation (19) instead of n we may 
put (n ± w), 

i.e. T (» ± m) = T t {n ± m + 1). 

T*_ 1 

T X+ 1 

n ± m 


— : i _ _ _ approx. 

i i 
= i — - 


n n'' 

In practice the coincidence may usually be limited to one of 
about 6, i.e. m = 3. 
A small value for n = 500. 

• ii-.i_.i-i 3 . 

T 500 ^ 5oo 2> 

i.e. an error of about 1 in 100000 is introduced, due to the 

In practice a cross-hair in the focal plane of eyepiece of the 
telescope is a useful reference point against which to estimate 

Sphere on a Coneave Mirror. An Approximate Method of Deter- 
mining *g', the Acceleration Due to Gravity 

A concave mirror is arranged horizontally, facing upwards, 
so that a small steel ball may be allowed to perform oscillations 
on its surface, in a line through the lowest point. 

The time of oscillation of the steel ball is obtained by timing 
as many oscillations as possible on the surface. The observation 
is repeated, and from these results a mean value of the periodic 
time T is calculated. 

Then if 

R is the radius of curvature of the upper face of the concave 

mirror, as measured by a spherometer, 
m the mass of the sphere, 
r its radius, 

g the acceleration due to gravity, 
it will be shown that 

T= 2, j5^- r ' (21) 

whence g = _? . !£. (R _ r ) (22) 

Consider the sphere in its position of equilibrium to be with 
its centre at B (fig. 37), and when displaced to the extreme 
position, with the centre at C. 


We will consider the case of a mirror of large radius of curvature 
and the displacement BC to be small compared with R. 

The potential energy of the sphere at C is mg • AB. Now 
AB = OB — OA = (R — f) (i - cos 6), i.e. the potential energy 


is (R — r) 2 sin 2 - • mg, or when is small as specified : 

P.E. = 2(R - r) (~fmg = \ (R - r ) 02 m S- 

The centre of gravity describes a circular path, BC in the 

diagram, so that ~ • = 0- 

R -r 

Hence the potential energy is 

i /T> . BC 2 i 

- (R - r) 75 i^wg = - 


(R-r) 2 

2R — r 

-•BC 2 . 

At this point there is no kinetic energy. 

At B the whole of the energy is kinetic, and equal to 

^mv m * + ^Iw m * 

where v m is the maximum linear velocity of the centre of gravity 
and w m the maximum angular velocity of rotation, and I is the 
moment of inertia of the ball about an axis through the centre of 
gravity, at right angles to the plane of the paper, i.e. 

K.E. at B is - mv m 2 + - 1 ?*£• 
2 2 r 2 

At any intermediate point P distant x cms. from B along the 


arc, the total energy is equal to either of these quantities and is 
therefore a constant, 

1 1 1 i x z 

i.e. - mx z H -x z + - mg-j=- . = constant, 

2 2 r 2 2 & (R — r) 

x being the velocity at that P along the path. 
Differentiating we have : 

mxx + - X 'X + (R _) ) * , * = °- 

Dividing by x and rearranging 

* = f- =-. x, 


i.e., the acceleration is a constant times the displacement ; 
this was shown (p. 25) to be S.H.M., whose periodic time T is 
given as under : 

|(R _,)(„+!_) 

T = 2TC = 27^ 

I mg y mg 

!(R - r) (m + I) 

Now, I, the moment of inertia about the axis described, is 
equal to — mr*. 

(R -r).Z 
Hence T = 2n W ^ (21) 


T is obtained by observations as indicated above ; r is measured 
by means of a screw gauge, R, by means of a spherometer, not 
by an optical method, unless the front surface is silvered. Hence 
all the terms in (21) are known except g, which may be 
calculated. The above method does not yield accurate values forg. 

Atwood's Machine 

A modern form of Atwood's machine is illustrated in fig, 38. 
The two masses, M x and M 2 , are equal, and are connected over the 
pulley, N, by a strip of white paper in the form of tape, while 
an equally long strip, M^Ma, connects the other ends of the 
weights, so that for all positions of M x and M a there is an equal 
mass of paper at each side of the pulley. 

The vibrator, V, carries an inked brush, B, and as^ the paper 
passes below it a trace, somewhat in the form of a sine curve, 
is drawn on the tape. The complete period of the vibrator 


is usually about one-fifth of a second ; but this should be carefully 
tested by means of a good stop-watch before beginning the 

The curve then provides a time record. The apparatus is 
used to provide an exercise in the determination of the accelera- 
tion due to gravity ; but even in its best form the experiment 
has no claim to great accuracy ; it does, however, provide an 
instructive exercise in Mechanics. 

Mj stands on a platform, as shown in the diagram, and the 
mechanism of the apparatus provides for the release of the 

Fig. 38 

vibrator and of M t simultaneously. Small weights are provided, 
which rest on the top of M x , and on release of the platform cause 
an acceleration of the masses. The rider can be removed by a 
second platform, P, after a velocity has been acquired. 

The trace on the tape records the acceleration and velocity 
beyond P, and from the former of these the value of g 
may be determined. 

There is always a frictional resistance to be accounted for, 
although this is reduced by making the pulley light and mounting 
it on ball-bearings. 

It is best to get rid of the retardation due to friction by making 


loops of wire, which may be placed on the top of Mi and remain 
when M 2 passes through P. 

The necessary addition can be judged approximately by 
observing the fall of Mj after it has been given a small velocity. 
If Mj moves down uniformly without appreciable loss of speed, 
the frictional error is nearly corrected. A finer observation may 
then be made by allowing the brush to make a trace. If the 
line consists of uniformly spaced waves the velocity is uniform. 
When this has been adjusted the wire loops are left in position 
and are not taken into account in the calculations. 

Suppose a rider of mass, m, lies on M x (in addition to the wire 

When Mj is allowed to fall, suppose it does so with an accelera- 
tion, /. 

Let Tj denote, the tension in the paper above M x . The lower 
paper strip is loose and is not supposed to exert any force on the 
two masses. 

Let I denote the moment of inertia of the pulley, and let T 2 

denote the tension in the paper on the left of the pulley, i.e. the 

tension acting on M 2 . Denote the radius of the pulley by a 

and its angular velocity at any instant by w. Then -j~ is the 

angular acceleration, and we have : 

dw . 

From the forces on (M x + m) we have : 

(m + MJf = (m + UJg - T x 
and from considering M 2 

Ma/ = T 2 — Mag, 
while the motion of the pulley is expressed by ; 

I^=(T 1 -T ! )a, 

from which we have : 

I/=(T 1 -T 2 )^. 
We may therefore eliminate T x and T 2 and find : 

j = g™ 

m + M x + M a + \ 

The quantity —is called the ' equivalent mass ' of the pulley, 

and its magnitude in grammes is engraved on the pulley. 

We may therefore find the value of g from this equation from 
observations which give the value of/. 


This may be determined from the trace. By 
removing P, the trace may be made long, M t 
being allowed an extended fall. 

The line drawn by the brush will consist of /^*B 
waves which open out uniformly. Mark these off in 
groups of five, as at A, B, C, etc., beginning at a 
point A, where the trace is opened out sufficiently 
to be distinct. Measure carefully the distances AC, 
BD, CE, etc., and divide by the time interval 
which elapses between these points. This will give 
the average velocity over the strips measured, 
and this velocity is the velocity at the points 
B, C, D, etc. The differences between these 
velocities should $all be the same, or very 
approximately so. Take the average of all the 
determinations and so obtain the average increase 
in velocity during five periods of the vibrator. Hence 
deduce the acceleration byj dividing by the time of 
five vibrations. 

This is the value of /. 

Repeat the experiment with the various riders ^q j 

Fig. 39 



All bodies, when acted upon by forces, are deformed a certain 
amount. The magnitude of the deformation produced by a 
definite applied force enables a value of the elastic constant of 
the material used to be calculated. 

We may, in a general manner, call the forces applied ' stresses,' 
and the deformations produced 'strains.' However, these two 
terms have, more often, a more precise meaning, depending on 
the mode of application of the forces. We shall recognize three 
ways of producing a deformation : (i) by uniform compression 
or extension, (2) by applying equal and opposite forces in one 
direction, i.e. stretching, (3) uniform shear. Deformation may 
be produced in any of these ways or by a combination of them. 

(1) Uniform Compression or Extension. 

If a body of volume V be subjected to a uniform pressure of 

p dynes per sq. cm., a contraction will ensue. This corresponds 

to a change in volume of <5V, say. The fractional increase in 

, . <5V 
volume is ^ • 

In this case the stress applied is p dynes per sq. cm!, and the 
strain is -= numerically. 

(2) Stretching. 

The most direct example of this type of deformation is seen 
in the case of a wire fixed at one end and supporting masses at 
the other end. In this case the force acting on the wire is the 
weight of the suspended masses and the reaction at the point 
of support. These are equal and opposite, acting in a direction 
which coincides with the length of the wire. Due to their 
action the wire will increase in length and at the same time will 
be reduced a very small amount in cross-section. The reduction 
in cross-section for a wire will not be of a sufficiently large amount 
to be readily measured directly. 

If 81 is the increase in length of a specimen whose original 

length is /, the fractional increase in length, the strain, is -=■• 




The stress producing this strain is denned as the force acting 
on each unit of area normal to it, i.e. if a is the area of cross- 
section of the wire and the total mass applied is m grammes, 

the stress is — - dynes per sq. cm. 

(3) Shear. 

Consider a cube of material ABCDEFGH (fig. 40) fixed 
at the lower face and acted upon by a tangential force F at the 
upper face. As a result of this force the cube will take up a 
position shown in an exaggerated manner by the broken lines 
in the figure, the vertical sides being sheared through an angle 
of <p radians from AH and BG to AH 1 and BG 1 . 

For such a shear — — „„ „ is the stress, i.e. the tangential 
force per unit area. 
The strain is measured by <p, i.e. the ratio 

EE 1 

if EE 1 is small 

compared with AE 1 . 

In all the above cases the stress is measured as a force per 
unit area ; in the c.g.s. system in dynes per square cm. The 
strain in each case is a ratio of like quantities, and has therefore 
no dimensions. 

Hooke's Law. 

If the stresses are below a certain limiting value which depends 
on the material of the body to which they are applied, the strain 
disappears when the stresses are removed. If the limiting value 
is exceeded, the material is strained beyond the elastic limit, 


and such strain is permanent ; as the stresses are still further 
increased the result is a fracture of the material. 

For stresses below the elastic limit, it was established by 
Hooke that the strain produced is proportional to the stress 
applied, i.e. under such conditions we have 


-7 — — = constant. 


The constant has a definite value which depends on the 
material, and which, in the three cases taken, is called, (i) the 
' bulk modulus of elasticity,' (2) ' Young's modulus,' and (3) the 
' modulus of rigidity.' 

Young's modulus is the most readily obtained directly by 

The following notation will be used throughout in dealing with 
these elastic constants. 

(1) Bulk modulus = K = £ 



(2) Young's modulus = Y = — = 





(3) modulus of Rigidity, n = — 

In addition to the above three elastic constants, we may add 
a fourth, which is concerned with stretching. We noticed that 
during stretching there is a lateral contraction of the specimen. 
The fractional lateral contraction produced is proportional to 
the longitudinal stress applied and the ratio of 

Fractional lateral contraction 

Fractional longitudinal extension 

is called ' Poisson's Ratio ' fp) . Thus, if the specimen is a cylinder 

of radius r and length I, and the changes produced in these 

dimensions are 6r and 61, we have 




The following relations between the elastic constants may be 



readily deduced (see for example, Poynting and Thomson's 
" Properties of Matter "). 

* -2( 3 K-f*) * (2) 

Thus, if any two of the constants are found experimentally, the 
remaining two may be calculated from the above equations. 

Determination of Young's Modulus for the Material of a Wire 

A direct method of finding Y is to support, vertically, a long 
length of the wire, load it with definite masses, and observe the 
extension produced. 

Fig. 41 

It will usually be most convenient to obtain such a length 
that, when supported at the ceiling, the wire extends almost to 
the floor. A second wire, C, is supported in like manner from 
the same support, and carries a fixed load of sufficient magnitude 
to keep the wire taut. The wire D carries a platform P. At V 
a vernier scale is attached to the wires, one half fixed to one 
wire, the other half to the second wire. 

The wire D, whose Young's modulus is to be determined, 
should be free from ' kinks ' and should carry sufficient load to 
make it taut, so that any further load added merely causes a 
stretching of the wire, and does not simply straighten out bends 
and kinks. If a heavy platform is employed at P, this weight 


may be sufficient. Read the vernier ; place one kilogramme on 
the platform and notice the extension. If this is due solely to 
the stretching of the taut wire, the vernier reading, on removing 
the kilogramme load, should be once more the same as at commence- 
ment ; if this is not so increase the load until on adding a further 
kilogramme the readings on the vernier scale have a definite value, 
which is reduced to another definite value on removing the load. 

Having obtained satisfactory repeats for this adjustment, the 
load on the scale pan P should be increased by equal increments 
and the vernier reading for each load noted. Having arrived 
at a safe maximum load, the latter is reduced by the same 
increments and the vernier readings again noted. 

The values obtained may be tabulated as under. 













mean extension 

for 6 kilos = 

The mean vernier reading for each load being taken and 
tabulated as shown, we may obtain several values of the extension 
of the wire for a definite load. 

Thus, in the case taken, the loads were o, 2, 4, 6, 8, 10, 12 kilos. 
The difference between the vernier readings for and 6 kilo load 
gives the extension for 6 kilos. In the same way the difference 
between the readings for 2 and 8 kilo load, 4 and 10, 6 and 12, 
also gives the extension for 6 kilogramme increase in load. This is 
entered in the last column, the mean value, I say, of which is 
used in the calculation of Y. 

The radius of the wire is measured in at least six places, using 
a micrometer screw, and the mean value taken, r cms. say. The 
original length of the wire, about 7 metres, is measured directly 
(L cms.). 

6000 x 981 




in the case taken. 



An alternative way (due to G. F. C. Searle) in which to measure 
the extension of the wire is illustrated in fig. 42. 

The standard wire terminates in a frame A which supports a 
mass M, sufficiently large to maintain the wire in a stretched 
state. The wire to be investigated is also fastened to a similar 
framework B. The two are fastened by cross-pieces C and D, 
which prevent relative rotation of the frames, but allow the 
frame B to be depressed relative to A, when masses are added 
to the scale pan S. 

Fig. 42 

A spirit-level L is supported at one end on a rigid cross-bar 
of the frame A, and at the other on the point of a micrometer 
screw V, which moves vertically through a rigid cross-bar. The 
micrometer screw has the usual circular division, which enables 
the movement of the head to be estimated to ts or rita of 
a complete turn, enabling the movement of the point of the 
screw (and hence the end of the spirit-level) to be measured to 
shs or ths of a millimetre. 

The level is first adjusted, when the wire is suitably stretched 
free from ' kinks/ so that it is truly horizontal. The load of, 


say, one kilogramme is added to the scale pan S. The micrometer 
screw is moved a suitable distance over to scale G, so that the 
spirit-level is once more horizontal. 

The amount of movement required to bring this about is 
obviously equal to the elongation of the wire by the load added. 

The results may be tabulated as in the former method and 
the value of Y calculated from the mean of a set of observations. 

Bending of Beams 

The value of Young's modulus may be found by less direct 
measurements for substances not in the form of a wire. 
Consider a rod of any uniform cross-section, say rectangular, 

Fig. 43 

bent into the form of a circular arc of fairly large radius. Take 
a section of the rod by a plane passing through the long axis of 
symmetry and parallel to it, and passing through the centre of 
curvature (i.e. the plane of bending). The layers of the material 
of the bar in the lower half will be compressed and the upper 
half extended. There will be one plane, therefore, at right angles 
to the plane of bending, which will remain of the same length as 
before the bending took place. This plane is called the neutral 
surface, and it will be shown to pass through the centre of gravity 
of the bar. It is represented in fig. 43 by NS. 

If we imagine the bar to be made up of a number of filaments 
along the length, then such filaments, as stated above, will be 
extended or compressed according to their position above or 
below the neutral surface. One such as shown at EF, fig. 43, 
or at P in the section diagram (fig. 44) a distance y above the 
neutral surface is extended. 



The strain in such a filament depends upon y, for, if R is the 
radius of curvature of the neutral surface, and the angle sub- 
tended at the centre O by NS, the unstretched length of the 
filament EF is the same as the present length of NS =R0. 
Also EF = (R + y)Q. 

Hence the elongation is yd and the strain is therefore ~L. 

Now if / is the force acting on the filament of cross-section a, 
we have from the definition of Young's modulus 






. / 


3 R 


i.e. fccy 

•O « SO Cf 





Fig- 44 

Thus, the arrows in the lower part of fig. 44 show the type of 
forces acting on all such filaments into which we have subdivided 
the bar. 

This system of forces on the bar must have an algebraic sum 
of zero. 



S/ = oor-^- Say = o. 

Since ^ is not zero, 

Say = o. 

Thus, the neutral surface passes through the centre of gravity 
of the bar. 

The forces have a definite moment about the neutral surface. 

The moment for the smgle force is fy = = .- ay 2 . 

For equilibrium the sum of such moments is equal and opposite 
to the external couple which set up the internal forces. If C is 
the external couple, we have 

Y „ Y . 

s r «r 



where * = Say 2 . From the similarity between this case and 
the corresponding sum in considering masses in connexion with 
moment of inertia, i is sometimes called ' the moment of inertia of 
cross-section.' It may be calculated in the same way as moment 
of inertia if area replaces mass. 
We have therefore for such a bent beam 

*Y = CR -....(3) 


Consider a light beam fixed horizontally at one end and loaded 
with a mass m at the other. If the mass of the beam is small 
compared with the load m, the whole depression may be taken 
as due to the load. 

In fig. 45 let AC 1 be the unloaded position for the neutral 
surface, and AC the position taken when the load is applied at C. 

To obtain an expression for the depression of the end in terms 
of the dimensions of the bar, m and Y, it is convenient to refer 
to a system of axes with the end A as origin ; AC 1 being the x 
axis, and a line at right angles to AC 1 from A in the plane of 
the paper the y axis. 

We will assume that the curvature of the loaded rod is small, 
i.e. the total depression at C is small. 

Consider a section at B, fig. 45, x cms. from A. As already 
seen, across the face of such a section a system of forces exist. 
These forces on the segment BC are extensions above the neutral 
surface, and compressions below, constituting a counter-clockwise 


couple C = -p on BC. 

At the same time the force mg at C has a clockwise moment 
equal to mg(l — x) on BC. 

For equuibrium these two opposite couples are equal in 

i.e: -^ = mg{l -x) (4) 


where R is the radius of curvature at B. 


Now R = jw- 



But in this case ^ is small — for the total depression is 

assumed to be small — so (~\ is negligible compared with 
unity, and therefore we have for such small curvature 




d 2 v W / x 1 

i.e. ^ = -W? (I — x) from equation (4) above. 

dx % *Y x 

The value of the total depression at the encf of the bar may 
be obtained by integration, and is the value of y when x = /, 
after such a process. 

Integrating we have 

BO -£( fc -3 (5) 

the constant of integration being zero, for when #=0,^=0. 

A second integration between the limits o and I gives 

L^Jo *Y L2 6 Jo 

where y is the end depression, 

i - e - y° iY 3 

For a bar of rectangular cross-section, i about the neutral axis 
NS (fig. 44) is — , where b is the breadth and d the depth of the 


12W I* 
Hence ^» = 6^Y*3 

v _4¥ 3 (6a) 

If the mass of the beam is not negligible, see treatment on 
page 94. 


Beam Supported at Two Knife-edges and Loaded in the Middle 

If the beam is now supported at the two ends and loaded 

with a mass m at the mid-point we have a reaction ^ at each 

knife-edge. If / is the total length of the bar, the depression 
produced at the centre will be exactly the same as the depression 

at the end of a similar bar of length - and loaded at the end with 


rrii .. 

a mass — , i.e. if we imagine the bar clamped at the mid-point 

and a force ^ applied at one end. The movement of the end 

is precisely the same as the depression in the middle in the 
actual case taken. 


Fig. 46 


Such a depression may be obtained by substituting the length 

- and force -£ in equation (6a), giving for the depression 

y ° ~ *Y ■' 48 
or Y==-^!_ ." >>(7) 

4bd 3 y 

The value of Young's modulus for the material of a beam may 
be obtained in this manner. 

The beam, say of iron, and of about 1 cm. square cross-section, 
and about 1 metre long is supported on two knife-edges near its 
ends, and a load is applied at the mid-point by placing masses 
on a pan which is suspended from a knife-edge which rests on 
the bar at this place. The depression is measured on a vernier 
scale, one scale of which is fixed, the other moves with the beam. 

The load is increased, and the vernier reading (y ) for each 

load is tabulated as under. The value for — is obtained in each 
case, the results being tabulated as under. 



load m 




2000 gms. 

3000 „ 

4000 „ 
5000 „ 




A mean value of — from the series of observations is taken. 

y Q 

The distance I between the knife-edges is measured directly, 
being of the order of 1 metre ; this can be done with a good 
degree of precision. 

Now d occurs in the third power, and is only a small quantity, 
therefore many observations must be taken and the mean value 
used. For d approx. 1 cm., an error of *i mm. means 1 per cent 
error, and this is magnified in d 3 to 3 per cent. 

Substitute the values found, in equation (7) 

^/m\ gl* 


Koenig's Method 

, Fig. 47 

Another method, due to Koenig, of determining the value of Y 
for the material of a beam, is by means of the type of apparatus 
shown in fig. 47. 


The bar carries a knife-edge which supports the load on a 
pan P. 

At the ends of the bar are two mirrors, M x and M 2 , almost 
normal to the bar, but slightly displaced to enable a scale S to 
be seen in the telescope T, the light from S having suffered two 

The telescope carries a cross-hair in the eyepiece, and the 
apparatus is arranged so that a scale division, as seen in the 
telescope, coincides with the cross-hair. 

If now the bar is loaded with, say, i or 2 kilogrammes, the 
mirrors will be turned towards each other as a result of the 
depression produced, and the scale division viewed in the telescope 
will be altered* This difference is noted = x divisions, say. 

Then we will show that 

Y ~ WdH ••••• (8) 

where W = Mg, M being the load in grms., 

I = distance between knife-edges, 
D r= distance between scale and the more remote 

mirror, M 8 , 
m = the distance between the mirrors, 
b and d having the same values as before, b 
the breadth, d the depth of the bar. 

For a bent cantilever we saw, equation (5) 
or between limits o and I, 


Now -2- is tangent of the angle through v/hich the beam 

has been bent. Let 9 be this angle. 

ti, W l % 

Then tan <p — -=- T 

»Y 2 

Now for in the present case, I being the whole length of the 
bar supported by two knife-edges, we obtain the angle through 

/ W 

which each end is turned by substituting V for I and — for W in 

the last equation, 

i.e. in the present case 


tan * = i67Y 

dxj i Y 2 



For rectangular bar i = 



tan 9 — 



For small depressions the angle <p is very small and so 
tan <p = q>. 

<p = 



Now a value of <p may be obtained from a consideration of 
the movement of the mirrors. 

Let m x and m t be the original positions of the mirror (fig. 48). 
In the first case the image of the division at D is in coincidence 
with the cross-hair ; when the mirrors move, each through an 
angle <p, F is then seen in coincidence with the cross-hair. 

Fig. 48 

Let us imagine the rays of light reversed, ABCD being the 
original path : when m x moves through q> and takes position 
m\, BC is moved through 2q> striking w 2 at E. 

Obviously m being the distance between the mirrors CE = 
m • 2q> very nearly. 

EG is a line drawn parallel to CD. The ray EF is swung round 
through an angle 49, for in addition to BE having moved through 
2<p, m 2 itself has rotated through <p. 
i.e. since GEF = 49 FG = 49* D, very nearly. 

But DG = CE = 2mp. 

;. x = DF = 2<p{m + 2D) 

<p = 

2(m + 2D) 



from (9) and (10) 

x 3W/ 2 

2(w -f 2D) 4&i 3 Y 

2&i 3 # 

y== 3 W/ a (m + 2D) ^ 

T P W 

In performing this experiment a mean value of — is obtained 

as in the last experiment, from observation of x corresponding 
to several loads and the other terms measured as before. Hence 
by substituting in (8) Y is obtained. 

Determination of Young's Modulus of the Material of a Bar by the 
Vibration Method 

We have already seen that the depression due to a load W 
at one end of a beam which is rigidly fixed at the other end is 
given by equation (6), page 89. 

This strain sets up an equilibrating internal stress equal to 


* L fs—y which acts as a restoring force. 

If the beam be allowed to oscillate in a vertical plane the 
restoring force when depressed y cms. is therefore 


Expressing the load as W dynes (m grms.) we may write, if y 


is the acceleration of the mass m = — at the end of the bar, 


W .. 3*Y 


ovy = -wi* > y 

This is a periodic motion, whose period T is 



T -">fe£- <"> 

or, as the mass at the end of the rod is m grammes, 

\ nd r 


A more complete treatment of this case may be seen under. 
In fig. 49 let O be the point of support, OA a vertical section of 
the unloaded beam, and OB the section of the beam when 
depressed by a load m at the end, so that the end depression 
AB = z , and the depression of the centre of gravity shown 


is u , whereas any point P on the bar is depressed a distance # , 
P being a distance s from O ', the total length of the bar is 
I cms. , 

Now let us imagine the bar to be homogeneous, and of mass 
6 per unit length ; so that its total mass is M = 13. 

We have, if i is the 'moment of inertia of cross-section,' 
considering the length (I — s), following the treatment of page 88, 

tY = R[W(Z - $) + (I - s)<5-g^^]|, 
taking into account the weight of the section of the bar concerned, 

Fig. 49 

which acts through the C.G. of the section. Since — = — 

if the beam is not greatly bent, 

,-Y g- = W(Z - s) +~(l - s)» 6-g, 
whence integrating 

for when s = o, x = o, i.e. the constant of integration is zero. 
Integrating between the limits s = o, s = /, x = o, to x = z 

=,(1^ ............. ...« I2 ) 

since M = d'l. 

This deals with a case of a steady depression. When the bar 
vibrates it is depressed below the position OB to some extreme 
position OC. To arrive at a value of the forces acting in the 
material of the bar in such a case let us assume that Wis increased 
to W\ causing the bar to be depressed into the position OC, 
where the extra depression of the end is z, of the centre of gravity 
of the projecting bar u, and of the point P is x. 


XT -V 73 W . M ^ & 

Now tYz =l 3 \- o~ 

3 8 

and iY^ + ^/'W^M^ 

3 o 

i.e. »Y* = — (W 1 - W) 

*=5^(W»-W). (13) 

In this case W — W is just balanced by the internal stresses 
in the beam which are thus equal to 

(W*-W)==-^s (14) 

This gives the value of the restoring force, hence the equation 


of motion of the mass — 'neglecting the mass of the beam, is 

W dH__ 3*Y 
g dt* I 3 '* 

or * = -&*S.z 

or Z*W 

X - e - Ni**" ^5) 

This is identical with the result expressed in equation (n), 
which was deduced from very simple considerations. It does 
not, however, take into account the mass of the beam itself. 
This is most conveniently done by considering the energy of the 

As the beam oscillates about OB, the steady deflected position 
of equilibrium, we will express the potential energy of the parts, 
using OB as the reference position of zero potential energy. 

Thus, in the position OC the Potential Energy of the system is 

— VJz — Mgu + (Energy stored as strain in the beam). 

The last term is obviously equal to the work done in straining 
the beam, i.e. is 

/ (straining force) dz 
(W 1 — W)dz = I ^tj — dz by equation (14) above 
3*X* 8 

* 2l* 


.-. P.E. is 

3*Y a_ W 2_ M gw _ ^6) 

2 I 3 

If instead of obtaining equation (12) by integrating the previous 
equation between the limits o and I we had integrated between 

the limits o and - . we should have obtained the depression u 

of the centre of gravity as under. 

*° *Y 48* +384 *Y 
. u may be obtained in a similar way to z (p. 96) to be 

u = 

W '- W ._L_Z3 

and since by (13) 

»Y 48 

So that in similar terms (16) becomes 

1 75-** - Ws - f 6 Mgz (17) 

Now the kinetic energy of the system is the sum of the K.E. 

of the mass-— I = - — £*) and the total K.E. of the vibrating 
g V 2 g / 


To obtain the K.E. of the beam consider an element of length 
ds at P. The K.E. of this element is 

£ d.ds(x) a 

x may be obtained by precisely the same method as z and u. 

* = ,4(t-iD(") 

so that x = -\---} z 

The total K.E. of the bar is 

2 v ' 



2 / 6 L V20 36 ^ 252/J 

280 280 
Thus the tote/ Kinetic Energy in the system is 

iT 1 '-'-^ 11 ** (l8) 

The sum of the Potential and Kinetic Energy is constant. 
Thus, adding (17) and (18) we obtain 

I/W + 33_ M Ua + 3*Y 2 _ m _ 4 M^ = constant. 

2\g I40 / 2l 3 16 & 

I /W + _33 M \ 2i g + 3 *Y 2 ^ _ Wi _ 5 M q 
2\g 140 / 2 I s 16 

Dividing by £ we have 

(7 + ^ M )'" + 3 , J«-(w + 4lfe)-o....(i 9 ) 

This represents a harmonic medium whose period T is un- 
affected by the constant term W + -4 Mg as seen below.* The 
period is the same as that of 

\g 14° / ' 

or T 

-«W V « 3( y° ; •; (20) 

* We may write equation (19) : 

put * = *-^( w + i! M s) 

/. 'p = Z, "P=" z 

= 2*^1 

W + -33_m > )/» 


Now, i, the ' moment of inertia of cross-section,' depends, of 

bd 3 
course, on the form of the rod. For a rectangular rod, i — — 

(about the neutral surface), where b is the breadth and d the 
depth of the cross-section in cms. 

The Experiment may very well be carried out using an ordinary 
boxwood metre rule. 

A definite length, /, of the rod is projected from the top of a 
table, to which it is rigidly clamped, as seen in fig. 50. 

f /////////y///////y/yyy^/y/^/^y^ 

Fig. 50 


A mass — is rigidly attached to the end of the metre scale, 

so that it has no 'play/ i.e. there is only the one vibration, 


namely, that of the scale itself. The value of — should be such 


as to cause but small depression. The rod is made to vibrate, 
and the time T is obtained by timing 50 vibrations in the usual 
manner. The length of the vibrating rod is altered, and the 
periodic time again determined. This is carried out for several 


lengths of rod, and also for several masses — at the end of the 



It must be remembered that M in the expression (20) for T is the 
mass of the vibrating part of the scale, and must be obtained for 
each length / employed. 

Of course, if m' is the total mass of the scale and L the total 


length, then M = -y- for the uniform rod. 

The results of the various experiments may be tabulated as 





1 ' 


W + ,33 M 
8 140 




11 (YL 
T 2 U 


_ i6it a (/ 8 /W 

+ 33M\ 
g 140/ 




is the mass attached in grammes. 

The method works equally well for such substances as wood, 
where the correction for the mass of the beam is small, and for 

brass, etc., where— M is comparable with the values of W. 


The modulus of rigidity is determined by observation of the 
twist produced in a wire by a definite couple, either statically, 
or less directly by torsional oscillation. 


Measurement of the Rigidity of a Wire by the Static Method 

Let us consider a long wire fixed rigidly at the upper end and 
subjected to a couple, C, turning it in the direction shown in 

fig- 5i. . 

We may consider a section of the wire (fig. 52), and of this 
section an annular ring as seen. Consider a small rectangular 
segment EFGHIJKL in the undisturbed wire. When the couple 
acts, this rectangular parallelepiped will shear as we have pre- 
viously seen, the faces making an angle <p with the original 
vertical direction. This happens throughout the ring, and 
adjacent sections will behave in the same way throughout the 
length of the wire. 

Fig. 52 

A vertical line, MN, along the wire (fig. 51) is moved by each 
ring through an angle <p, and so finally takes up a position MP 
when MP is inclined at an angle g> to MN. 

If I is the length of wire from the fixed end, and R is the radius 
of the wire, then since N is moved to P when the couple acts, 
the radius ON moving to OP, where the angle NOP =6 radians, 
we have 

arc NP = h = R0 • • --(21) 

Considering again the annular ring of fig. 52, of radius r and 
width &r, we have the shear produced by the couple C. 

Let /be the value of the shearing force over the area, 2-nr-Sr, 
of its upper and lower surfaces. From the definition of rigidity n 


2-nr ' dr 
n == «^— 

or f = 2isnqrrdr. 

Since this force has a moment fr about the axis of the wire, 
we have, replacing <p by the more easily measured term 0, from (21), 

' 2w»0 

ft — T 


The total couple throughout the solid section is therefore : 

Jo Jo I 2l 

a couple which is equal to the applied couple C for equilibrium, 

C-=£-' (») 

This may be re-written : 

C . ttR* 
-'I — • n. 


-, the couple required to produce unit angular twist, is called the 

' coefficient of torsion ' = t, say. 
We thus have : 





tZ = «-"-^- .(23) 

xl = i • n. 
=i the moment of inertia of cross-section for the circular wire, 

and in equation (22) is measured in radians. 

Fig. 53 

Experimental Details. 

The method of finding the coefficients is to fix a wire specimen 
rigidly at one end, apply a measured couple at the other, and 
measure the twist produced at a given distance I from the fixed 

The two types of apparatus usually employed are shown in 

figs. 53 and 54. 



Using the vertical wire type, pointers are fixed to the wires 
at different distances from the fixed end, and a couple is applied 
to the free end by adding weights to the scale pans S, S 1 . The 
amounts of twist, V a > 3 , at the distances, l lf Z a , Z 3 , from the 
clamped ends are observed on the circular scales shown. It will 

f) ft f) 
be found that -~ = ~ = -~ as we may expect from equation (22). 
L x / 2 / 3 

Now consider a fixed length of wire I cms. from the support. 
If M is the sum of the masses applied in S and S', and the diameter 


of the wheel at which the couple is applied is D, then Mg— = C 

in equation (22). 


A series of values for 0, corresponding to various masses on 
the pans, is obtained. These results may be tabulated as under. 
























The last column of the table shows the ratio — to be a con- 

stant — or nearly so. For such a variation as the one shown in 

the table of results above, the mean value of — is taken and 

substituted in equation (24). Re-writing (22) we have 

„ D TmR* B . „. 

M-g • — = — j — 6 e in radians, 

M ,gV>l 
or n = —• • -s— 6 in radians. 

tcR 4 

The length of the wire is measured directly. The arm of the 
couple, D, is measured by means of callipers. 

As R, the radius of the wire, occurs in the fourth power, especial 

care is taken to obtain its true value. Determinations of R 

are made with a micrometer screw gauge at several points along 

the length of the wire, and a mean value taken. 

^ M 

Substituting the mean value of — from the table of results, 

all the unknowns of the equation are ascertained. However, 

as is usually measured on the circular scale in degrees, the 

value of the ratio — must first be converted to radians. 

Assuming therefore now that is measured in degrees, the 
end result which converts to radians, etc., is 

M I g(i8o)D , x 

The same type of observations are necessary with the horizontal 
apparatus, which makes use of a shorter length of wire specimen. 

Maxwell's Needle 

If a bar, AB, is suspended horizontally by means of a wire 
whose modulus of rigidity n is to be determined, the value of n 
may be obtained in terms of I, the moment of inertia of the 
bar about the axis of suspension, and T, the time taken for the 
bar to make a complete horizontal oscillation. 

Consider a small displacement from the position of rest of 
the bar. We have already seen (p. 102), that in such a case the 
couple called into play in the wire to equilibrate the displacing 

couple is t = -j- per unit angular displacement, where 

tcR 4 

i = , so that the couple exerted by the wire for an angular 

displacement 6 is t0 for that particular wire. 



It has been shown (p. 53) that the moment of such external 
forces on the oscillating bar is 

T dH 


I '-!£- - rS 


i.e. these couples are equal and opposite at any point. 

^ = "I 
This will cause vibrations whose periodic time, T, is given 
by (p. 25) 

Hence, knowing T by timing a number of swings, if I can be , 
ascertained t may be evaluated and hence n determined since 
*R 4 n 

Fig. 55 

/ In Maxwell's needle (fig. 55) the bar is replaced by a hollow 
tube, of length D cms. Fitting in the tube are four equal-sized 

cylinders, each of length/ — j, two of solid brass, and two 


Let I„ be the moment of inertia of the hollow cylinder of 

length D, about the wire as axis, 

I ! be the moment of inertia of the solid brass cylinder 

about a parallel axis through, its centre of gravity, 

1 2 be the similar moment of inertia for the hollow brass 

tn x the mass of each solid brass cylinder, 
m 2 the mass of each short hollow brass cylinder, 
I the length of the wire in cms. 
In the first case place the cylinders in the order shown in 


fi g- # 55 (a) and find the periodic time T x . Then arrange the 
cylinders as in (b), once more obtaining the periodic time 
by timing, say, 50 complete swings. Let this time be T 2 . 

Suppose that the moment of inertia of the complex bar in 
the first case is I', and I" in the second. 

We have T t = 2* JX 

T 2 = 2tc <J — 

Ti* - t,« =^-Vr -r) 

Now we have 

for the moment of inertia of each of the four units is given by 
the ' law of parallel axes,' 

I' = I + 2l 2 + 2 m 2 (^ + ali + zm t (^) 2 

n D 2 D 2 / \ D 2 

= 2w x -g - 2w a -g- = f m x — m 2 J — 

By such an arrangement I' — Y may be evaluated and hence 
t may be calculated. 

To enable an accurate measurement of T in both cases, the 
hollow frame carries a small mirror. A beam of light from a 
lamp is focussed on the mirror, and the reflected beam is directed 
on a scale. As the reflected beam, shown as a spot of light on 
the scale, passes a certain mark in one direction, a stop-clock is 
started, and, starting counting, 1, 2, 3, etc., as the spot again 
passes the scale reading in the same direction, the time for fifty 
complete oscillations is obtained, when the four short cylinders 
are arranged as in fig. 55 (a) and (b), whence T t and T 2 are 
obtained. The length I of the wire from the rigid support, 
and the length D are measured directly : m x and m 2 are obtained 
by weighing, and therefore all the factors for the determination 
of n are measured, for, 

2 I' — F . .m, - m, D 2 

and n 

4w 'TV=Tr» = 4w TV 

_ 2tZ 



Determination of the Modulus of Rigidity of the Material of a Flat 
Spiral Spring 

In this experiment the value of n for the material of the wire 
of a flat spiral spring is deduced from a knowledge of the periodic 
time of vertical oscillations of the spring when loaded with a 
known mass, and the dimensions of the spring. 

Let us consider a flat spiral spring, whose turns are closely 
wound with a wire whose radius r is small compared with the 
radius of the spring itself ; such a spring may be made by winding 
the wire on a wooden cylinder of suitable diameter. 

Let the ends of the spring be bent twice at right angles, the 
free ends, such as BC (fig. 56), being therefore along the axis of 
the cylinder. 

If the spring is clamped vertically at one end, and loaded with a 
mass, M, at the other end, as shown in the figure, the force, Mg, 
along the axis exerts a couple tending to twist the wire in the 
direction of the arrow. 

It was shown* on p. 102 that, under such circumstances, if I is 
the length of the wire from the fixed end, and R the radius of 
the wire, 

C = f^ 6 (25) 

2 I 

In the case of the spring we may apply this formula when / 
represents the total length of wire, and is equal to 2ttRN, where 
N is the number of turns. 
If r = radius of the wire, 

R = radius of the cylinder on which the spring is 

wound -f «r, 
I = 2«RN, 
M = the load in grammes, 

to c -'r-iaar (25a) 


Take a section of the spring at A, shown in fig. 57 enlarged. 
When the couple is applied the arm AB is twisted through the 
angle 0, given by equation (25a) above, and takes up the position 
AB 1 . 

Fig. 57 

The depression x, of the end B, is obviously equal to R0, 
approximately, when R is large, 
i.e.. * = R.***!C = 4NR!C 

If / be the restoring force on M due to the wire, the 
couple C =/R. 

ie - *5RF=' "• (26) 

The equation of motion of the moving mass is therefore 

M dt* ~ 4NR 8 ' *" 

This was shown (p. 25) to represent S.H.M., whose period is 
given under. 

T = ->/^ r w 

From observation of the periodic time of the spring oscillating 
in a vertical manner, n, the rigidity coefficient of the material 
of the spring can therefore be obtained. 

However, in the foregoing we have neglected the mass of the 
spring itself, and have also assumed that the total depression 
produced by M was due to twist alone. 

We must now consider these factors. 

Shear in the Spring 

We can very easily see that the shear effect in the type of spring 
taken above is of a negligible order when the radius of the 
wire is small compared with the radius of the spring. For this 
purpose again neglect the mass of the spring. 

If / is the force acting along the axis, at any point A in 
the spring there is an equal and opposite force,/, for equuibrium. 

Thus, for the small segment AD, shown enlarged in fig. 58, 
at A there is a downward force /, and at D an upward force /, 
constituting a shearing couple which would cause a depression 



of A to some lower position A 1 , through an angle 9', say, where 
from the definition of rigidity 

v An 
The area A is the area over which the forces / act, which, in 
this case, is the area of cross-section of the wire, nr 2 , 

„> - f 



Fig. 58 

The total depression due to this shear for a length / of the wire 


Now, using the expression for the depression due to twist, 
given in equation (26), and substituting I = 2«RN in the ex- 
pression (28) above, we have 


Depression due to shear _ nrhi r^ 

Depression due to twist /4NR 8 ~~ 2R a 

Take an average case, r = -05 cm., R = -8 cm. 

r* = 1 
2R* 512 
i.e. depression due to shear is about -2 per cent of the depression 
due to twist. 


Thus, when r is small compared with R we may neglect the 
shearing effect. 

The Mass of the Spring 

The expression for the time of vibration is modified if account 
be taken of the mass of the spring itself. 

The total mass moving is greater than M. We may regard 
the problem as being similar to the case of an ideal massless 
spring, loaded with (M + m'), where m' is an additional load 
which just has the same effect as the mass of the actual spring 
used ; m x may be called the ' equivalent mass of the spring ' and 

T = 2tt l (M + m') 4 NR 3 ~ 
\ r*n 

Thus, using this formula, we may find n and m' by obtaining 
T for two loads, and solving for m' and n. 

However, we may treat the problem in somewhat more detail, 
and deduce an expression for the equivalent mass m', in terms 
of the actual mass m of the spring. 

It has been shown that the twist of the wire of the spring at 

the end is = f— , since C, the couple, is equal to MgR. 

The depression of the end due to this twist is R0. 
If this depression is x, 

* = 2 ™^! (29) 

In the vertical oscillations, the mass M at the end of the 
spring is depressed further than this. To arrive at a value of 
the forces called into play due to such further straining of the 
spring, suppose a bigger mass, M 1 , be applied, causing an increased 
depression z, i.e. total x -{• z. 

Then . x + z = ~^- 

* = ^-(M*--M)*» (30) 

To allow for m, we may consider the energy of the system 
with regard to the position of rest of the loaded spring. 

•The internal forces equilibrating the force 

(M 1 — M)g, axe therefore = —r^i- 
Therefore the equation of motion of the mass M is 

Mz= — 

= 2 *Y 


as previously shown, equation (27), neglecting the mass m of the spring itself. 


When displaced a distance z from the position of rest (which 
is x cms. below the unloaded end) 

Potential Energy is 

—Mgz—mg (displacement of centre of gravity) + (energy stored 

as strains in the spring). 

The centre of gravity is lowered a distance which is equal to 

half the depression of the end= - . 

The energy of the spring is equal to the work done in straining 
the spring, i.e. since the internal forces are equal to (M 1 — M)g 

= — rp-'^i from (30), the work done in depressing the end a 

distance dz is l—p~\zdz, 

i.e. energy in the spring is 

•ttrhi , _ nr*n z 2 

2mF' z ' dz " iJR 2 ' 2 

Total P.E. is therefore 

7tr% 1 

___*2_Mg Z - -mgz. 

The Kinetic Energy is 

\ M(i) 2 + K.E. of the moving spring. 
If 6 is the mass of unit length of the spring wire, the second 
term above is 

/ — (mass of a small element) x (velocity of the element) 2 . 

At a point P s cms. from the point of support measured along 
the length of the spring, consider a small element ds. The 

velocity of that element is j • z, its mass is dds, 
i.e. K.E. of moving spring is 

. 2J 

dds • j 2 

= i*Ly-=I(™W since m-M. 

2 l z 3 2 \ 3/ 

The total energy of the system in the position considered, 

and which is constant by principle of conservation of energy, is 


^ -(^ + ^, + 1(11+5),.- constant. 

Differentiating and dividing by i we have 

5£.-(* + «.) + („ + !).-.; 

as shown on footnote, p. 98, such a system is a periodic motion 
whose periodic time T is the same as for 



2.1 t 

M + ™W 

i-e. T = 27r-JJi f (31) 

The effect of the mass of the spring is therefore the same as 
though M were at the end of a massless spring together with a 
load equal to | the mass of the spring. 

Experimental Details. 

A flat spiral spring is chosen, having a radius R which is 
fairly large compared with the radius of the wire. One end is 
clamped firmly in a heavy retort stand ; a mass is attached to 
the lower end of the spring, and the time of vibration of the 
vertical oscillations is obtained by timing 50 vibrations with a 
stop-watch. This is repeated for various loads. 

/, the length of the wire in the spring, may be obtained from 
a knowledge of N, the total number of turns in the spiral. If 
there are exactly N turns, the length is 2tcNR where R is the 
average value of several observations of the mean radius of 
the spiral, i.e. the value of the outside radius of the wire spring, 
minus the radius of the wire of which it is made. 

r occurs in the fourth power, so should be measured with 
extreme care. A number of values are obtained with a screw 
gauge at points along the length of the spring, and the mean 
value taken. 

/ being equal to 2tcRN, we have from equation (31) 

i6k 2 R 8 N + 3 .. 

n = — jT-'—fr* - (32 ) 

The results of the several experiments may be conveniently 
tabulated as over. 




(M + tn/3.) 



M + w/3. 


mean value of 

M H — 

The mean value of 

— ^ for the series of loads (M) taken 

is obtained and substituted in equation (32), thus giving the 
mean value of n for the complete set of observations. 

Determination of Young's Modulus of the Material of a Spring 

The value of Young's modulus for the material of the wire 
of which a spring is made may be obtained by supporting the 
spring vertically and allowing a bar, which is firmly fastened 
to the lower end, to perform horizontal swings as in the case of 
Maxwell's needle (p. 102). 

It is shown below that in such a case the time of a complete 
horizontal oscillation is 





is Young's modulus for the wire, 

is the radius of, the spiral, measured from the centre 

of the wire, 
the radius of the wire, 
the moment of inertia of the bar about the axis of 

N the total number of turns in the spring. 

The above assumes, as in the last experiment, that the spring 
is a flat spiral ; the layers are very close together, and each may 
therefore be regarded as horizontal. 

Let fig. 59 represent a horizontal section of a small length s 
of the spring, with NS the section of the neutral surface by the 
plane of the diagram. This neutral surface will be normal to 
the diagram. ABCD is the section of the element of the spring 
when the bar is in an undisturbed position. 



When the bar is turned in a horizontal plane through any 
angle y>, the section taken, in common with every other element 
of the spring, will become more curved, as shown in the broken 

Consider a single ' filament ' of the material, as was done in 
the case of the bending of beams (p. 86). If a is the area of 
cross-section of the filament, and / is the force required to bend 
up the filament through a small angle 

Y = 

Strain of the filament 

Now, if R is the normal radius of curvature of the neutral 
surface, and PQ is a distance x cms. beyond NS, the original 

Fig. 59 

length of PQ = (R + x)6 , where O is the angle between the 
radii from the original centre -of curvature O to the ends AC 
and BD of the element considered. 

Now, due to stresses similar to /, the element has a greater 
curvature, and consequently the neutral surface has a smaller 
radius of curvature R. If O 1 is the new centre of curvature, 
the new length of PQ is (R -f x) 0. 

However, the neutral surface is Of the same length as before, 
so that R O O = R0. 
The strain is therefore (R + x)S — (R + x)d -f- original length 

== — — f — — approximately. 

S = Rn0fl = R0- 


Strain is therefore 

* \R0 "" *W = * \S ~ Ro/ 


• Y = 

The moment of this force on the filament about the neutral 
surface is fx. The total moment of the couple acting on the 
element considered is therefore 


If the angle subtended at the centre by i cm. of the wire 
be q> and <p Q , respectively, the above couple is 

"C = Yi{v - <Po)- 

If I is the length of the wire in the spring, the total change 
of angle, i.e. the angle through which the inertia bar moves is 

y, = l(<p — <p Q ) 

i.e. C-Yt*- 

This is the value of the- couple due to the internal stresses, 
called into play by the strains in the wire, 

i.e. Iv = — j- v 

The periodic time T is given by 

i = ^1, 1 = 2*RN 



T = aTc^J^Y - 

1 ~" r*T* 


Experimental Details. 

The spring is set up as in the last experiment, and a rod, say, 
of rectangular cross-section is clamped to the end of it, so that 
there is no free play between the end of the spring and the rod, 
and the centre of gravity of the rod is under the suspension. 

The ' inertia rod ' is then given a displacement in the horizontal 
plane, and the subsequent horizontal oscillations are timed. T is 
obtained by timing 50 complete swings in the usual way. 

r and R are carefully measured as previously described, and 
the total number of turns of wire in the spring, N, is counted. 

The value of I may be calculated from a knowledge of the 
dimensions and the mass of the bar (see p. 54). 

Determination of the Modulus of Rigidity and Young's Modulus for 
the Material of a Wire by Searle's Method 

To find Y, Young's modulus, or n, the coefficient of rigidity, 
of the material of a wire specimen by this method, the wire is 

Fig. 60 

fastened to two identical rods, &J& lt A 2 B 2 , at their mid-points, 
as shown in the diagram at C^ C 2 . These rods are usually 
square or circular in cross-section, and are supported by threads 
from points T 1 and T 2 , such that the axis of suspension and the 
axis of the wire intersect at the centre of gravity, and the sus- 
pended rods are a distance apart, which allows the wire to be 
stretched in a straight line, the whole assuming an H formation. 

If now the ends, B x and B 2 , of the rods are drawn together 
and fastened by a loop of thin cotton, the wire will be bent into 
the arc of a circle, subtending an angle 2<p at the centre ; each 
rod will make an angle q> with its original direction. 

The suspension of the two rods being such that the torsion 
is negligible, the only couple acting on the bars is that due to 


the bending of the specimen, CjCj. With this method of suspension 

we have a couple produced equal to ^ in the wire (p. 88), 

where * is the ' moment of inertia of cross section/ and R is the 
radius of curvature of the arc C^. 

Fig. 61 

Now, if I is the length of the wire between the rods, 



If r is the radius of the wire (assumed circular), 

• = *^ 

4 ' 
whence the couple acting is 

4 * I ~~ 2/ 
Let I x be the moment of inertia of the rod, kj& x , about the 

axis of suspension, -57J the angular acceleration. The moment 

of the external force is, by the theorem given on p. 53, 1-j^- 

This is equal and opposite to the restoring couple exerted by 
the bent wire. The equation of motion for the rod is therefore 

This is simple harmonic motion (p. 25), and 

T =2K J5T 


If the rod AjBj has a length 2L and width 2a and mass 
M grammes, 


L 2 + a* 


Y = 

8tt/ L 2 + fl2 


y 4 T 2 3 

To evaluate Y for a specimen of wire, the arrangement 
described is fitted up. The thin cotton loop holding the rods 
together at B X B 2 is burnt, and the resulting oscillations are 
timed with a stop-watch, T being obtained by timing 50 complete 
oscillations. An alternative method of finding T, and which 
may be used in many experiments where T is required, is seen 

A reference point having been chosen — say, a chalk -mark 
under the oscillating bar — the time is noted at which the counting 
of the swing is commenced. After 5 swings the time is again 
noted, and the times at the end of every 5 is recorded in a table 
as seen below. 














a — u 





b -V 





c — w 





d — x 

20 sm 





e -y 

mean time for 30 oscillations. = 

Tabulating in the second column the time for the starting point, 
5, 10, 15, 20, 25 oscillations, and in the fourth column the time 
for the 30, 35, 40, 45, 50, 55 oscillations, the difference in any 
line between the fourth and second column value will give the 
time for 30 complete oscillations. The mean value of the last 
column being taken as the time for 30, T may be evaluated. 

The radius of the wire occurs in the fourth power. The value 
of the diameter is therefore measured in several places, say, six, 
and the mean radius, r, calculated. The other measurement for 
the evaluation of Y are straightforward. 

It will be noticed that in this method the value of Y is obtained 
by timing, and not by observing deflections as in some of the 


other methods. Another advantage of the method is that it 
requires only a small specimen of the material. 

To find the modulus of rigidity of the specimen, the rod A 2 B 2 
is clamped rigidly in a horizontal plane and the wire acts as a 
support for the rod AjBj, which may therefore be made to 
perform horizontal torsional oscillations. The rod is displaced 
slightly from its position of rest causing a restoring couple to 
act on it, due to the twist of the wire. 

For a displacement 0, the restoring couple is 

as established on p. 102. 

We therefore have as the equation of motion of the rod 
T d 2 d _ _ nnr 4 * 
dt* ~~ 2.1 
This is once more a case of simple harmonic motion, whose 
periodic time, Tj, is given by 

\ = ^ 


If the rod is of square or circular cross-section, the moment 
of inertia about the axis taken in this experiment is the same 
as the last case. If of rectangular section, the moment of inertia 
will be obtained by calculation as before. 

Tj may be found as described for the first experiment; the 
length I and the radius r are already known. 

Hence n may be found by substitution in the formula below : 

_ 8tc/I 

Determination of the Bulk Modulus for Glass 

The bulk modulus, as shown on page 82, may be expressed as 



where p is an increase in pressure causing the small volume 
change 6V. 

In general, it will not be convenient to apply a uniform pressure 
to a body, but it is a simple matter in most cases to apply an 
extending force per unit area, or a pressure, in one direction 
only. If, as a result of such a pressure, the change in volume is 

<$V l , then 6V 1 — — • 6V where <5V is the change in volume when 

the same pressure is applied uniformly over the body. 
If a cylinder be supported at one end, and an extending force 


be applied to the other, in a direction coincident with the axis 
of the cylinder, we have the extending force at one end and the 
equal reaction at the other. The change in volume in this case 
is of the type 6V 1 above. So that if the change in volume of a 
body be observed in such a case, the value of k may be calculated. 
The apparatus used to determine the bulk modulus of, say, 
glass is seen in fig. 62. A glass tube g is cemented to two caps 
of brass, A and B. The upper brass cap A is provided with 
two pegs, which act as a support for the whole apparatus. The 
lower cap B carries a hook to which a pan S is attached. 





Fig. 62 

Fitting in the upper end of the tube is a rubber cork, carrying 
a capillary tube, CD, which is graduated and calibrated in 
the manner described on page 41. 

The tube g is filled with water. Care is taken to avoid 
trapping air, and the cork, etc., is placed in position, resulting 
in a little water rising in the capillary tube, i.e. the whole of the 
glass tube and part of the capillary are filled with water to a 
definite level. If now the pan S is loaded with, say, 5 kilogrammes 
the volume of the glass will increase by a small amount. The 
difference in the levels on the calibrated tube enables the value 
of the volume change to be determined if the temperature 
remains constant throughout the observation. 

However, the apparatus is, by the construction, very much 
affected by small temperature changes ; it is a water thermometer 
with a very large bulb. To obtain a good approximation to the 
value of 5V 1 , the volume change, the reading on the tube is 
noted when S has no load ; 5 kilos are applied and the reading 



noted, the load is removed, and once more the scale reading is 
taken and the process repeated with 5 *& os ^ no load for 
10 observations, each one being taken at a regular time interval 
as under. 











o min. 

o kilos. 

5 min. 

o kilos. 












7 • 


















A similar process is repeated with a load of io, 15, and 20 kilos. 
The results are plotted as a graph, using a fairly large scale 
for the ' scale readings ' as ordinates. 

O ^ Time in minute© . 

Fig. 63 

The no-load curve will be a continuous line, and the short 
curves for 5, 10, etc., load will be at a vertical distance above 
it, depending on the load employed. 

The form of curve obtained in a particular experiment, using 
one type of glass, is seen in fig. 63. 

AA X , BB 1 , CC 1 are then drawn parallel to the axis of scale 
readings, and their values measured : the mean value of this 


length gives the scale reading increase in volume of the glass 
tube g. The actual volume change is then obtained from the 
calibration curve of the capillary tube. 

This process is repeated for the curves giving the results for 
each load, the mean of the three ordinates being taken in 
each case. 

From each value of <5V the increase in volume SV per kilo 
load may be calculated. A mean value for all loads therefore 
gives the mean for all the observations taken, for the value 
of SV per kilo. 

The original volume of the glass tube may then be found by 
obtaining the product of the length between the brass caps and 
the area of cross-section. 

To obtain the cross-section, a little water is poured into the 
empty tube so that the surf ace becomes visible above the lower 
cap B. A measured volume of water is then run into the tube, 
so that the level is just below the upper cap. From a measure- 
ment of the increase in the level and the volume introduced 
from a graduated flask or measuring jar, the average cross-section 
may be calculated. Hence V is obtained. 

To find the stress applied to the tube we must find the internal 
and external radii of the tube. The load applied at the end 

causes a stress, a force per unit area, equal to /T -. m ^ — r-, where 

7U(R 2 — f 2 ) 

m is the mass, R the external and r the internal radius of the tube. 

Now, Ttr 2 was determined above, when obtaining V. To find 
R, the external radius is measured with callipers at about 6 or 8 
places, and the mean taken. 

Thus, if 6V is the mean value of the volume change per kilo- 
gramme load, i.e. for a force of 981 x 10 3 dynes in one direction, 
the change for a uniform stretching force would be $dv, 

981 x io 3 
and K = -r— — dynes/cm a . 

Poisson's Ratio for India-rubber 

A length of solid india-rubber is supported at one end and 
loaded at the other. The rubber, of. circular cross-section, is 
marked at about six places along its length, and the diameter 
at these selected places is measured with a micrometer screw 
for each load placed in the scale pan at the free end of the india- 


The length of the specimen is also measured for each load. 
Then, since Poisson's Ratio has been denned as 

Lateral contraction / 
_ Original diameter 
^ ~ Longitudinal extension 
,. Original length 

(i may be calculated for each load, the mean value from the six 
measurements in the lateral direction being used. 

For a rubber cord of about f-inch diameter, suitable loads 
would be 500, 1000, and 2000 grammes. 

It will be found that, to some extent, the values obtained 
depend on the history of the specimen. The results when the 
load is increasing will be somewhat different from the values for 
the same load when decreasing. 

In all observations the readings should not be taken until about 
ten minutes after the adjustment of the load. 



The surface of a liquid acts in many respects in a manner 
analogous to a stretched membrane. The well-known example 
of mercury resting on a clean wooden surface shows the effect 
to a marked degree. The mercury takes the form of a globule, 
as if it were surrounded by a membrane supporting it in this 

The examination of a water drop slowly formed at the end of 
a glass tube or tap from which it emerges provides another 
example of this phenomenon. The water in this case accumulates, 
as though it were collected in an invisible membrane, until of 
a definite size, when it is detached as a spherical drop. 

D i B 



Fig. 64 


These effects are due to forces existing in the surface of separa- 
tion of the liquid from the air and the other media in contact 
with it. The effect is generally known as surface tension. This 
term may be defined in two ways, depending on the point of 
view taken. 

If we imagine the surface of the liquid to be cut by a plane, 
there is a definite force per cm. acting on the line of intersection, 
at right angles to its length and parallel to the surface. This 
force, expressed as dynes per cm., is defined as the surface 
tension of the liquid. The value of this force depends on the 
liquid and the surrounding medium, but, unless otherwise stated, 
will be taken throughout this book to represent the force per 
cm. when air is the medium. 

For example, consider a frame ADEB, fig. 64, of width I cms., 
across which a film of liquid, whose surface tension is T dynes 
per cm., is stretched. 

If the film terminates at the lower end on a light rod AB, 



since the liquid film has two surfaces, it will exert an upward 
force of 2T/ dynes on the rod. If the rod has a mass of m grammes 
it will be in equilibrium when 

217 = mg. 

Now, suppose the rod be displaced a small distance 8x, to A*B 4 , 
against the surface tension forces, the work done is 2TI • dx, 
and the resulting increase in area of the surface is 2I • Sx. 

Thus, the work done per sq. cm. of surface is 

2Tl6x „ 

which may be taken as defining surface tension, i.e. the surface 
tension is the work done in enlarging the surface by one cm?" 

When a liquid is placed on a horizontal plane surface, the 
form it takes depends, for a given liquid, on the material of which 
the plane surface is made. Thus, if water is placed on a clean 
glass surface it spreads over it, whereas if the glass is greasy 
the water takes the form of globules. 


u v-— < 

1 • 


Fig. 65 

The angle contained between the plane surface and the liquid 
surface is different in each case. If we measure this angle in 
the liquid we have a measure of the angle of contact. Thus, in 
fig. 75, showing a section of a mercury drop on a glass surface, 
is the angle of contact, whereas for a liquid like water which 
' wets ' the glass the angle of contact is zero — the water spreads 
over the surface. 



(1) Wilhelmy's Method 

To determine the approximate value of the surface tension 
of such a liquid as water, paraffin oil, or turpentine, the following 
method may be employed. 

A clean wire, preferably platinum, is bent into the form shown 
in fig. 65, making three sides of a rectangle of breadth / cms. 


It is then suspended from a beam of a balance by means of 
a thin wire, and counterpoised when the upper horizontal arm, B, 
of the frame is almost immersed in the liquid. 

When a balance is obtained the frame is dipped under the 
surface of the liquid by lowering the beam of the balance, and 
then withdrawn again. A film of the liquid will be formed in 
the frame as seen in the shaded part of fig. 65. 

Due to the downward surface tension pull on each side of the 
film of length I there will be an apparent increase in weight = 2IT. 
This can be found experimentally by adding ' weights ' to the 
other scale pan> until on raising the beam it remains horizontal. 

Fig. 66 

To perform this approximate experiment clean the frame by 
holding it in a Bunsen flame until red hot. 

Repeat the experiment as described above several times, and 
take the mean value, whence if m is the mean value of the added 

2T/ = mg, 

T = — j- dynes per cm» 

Take care that the frame is the same height above the water 
surface when the balance is made, before and after immersing 
in water, to eliminate buoyancy errors. The surface tension 
effect on the two vertical limbs is eliminated, but the experi- 
ment may be regarded as one which gives an approximate value 
of T as described above. 

(2) By Weighing Drops 

The liquid whose surface tension is to be measured is allowed 
to form drops at the end of a narrow tube, C, fig. 67. If m is the 
mass of the drop, we have 

mg = KT, 
where K is a constant. 



From a simple approximate investigation of the case we see 
that mg = 7tfT as under. 

When the drop is about to break away from the tube we will 
assume it has the cylindrical form shown in the firm lines, D. 
The broken line indicates one of the subsequent forms. 

If r is the radius of the orifice, we have, at this level inside 

the drop, an excess of pressure equal to — due to the cylindrical 

curvature of the liquid surface. This is equivalent to a downward 

force — • T?r 2 =Tnr. 

J y 

The weight of the drop being mg, the total downward force is 

T7rr + mg. 

This is equal to the upward surface tension force over the 
circle of contact, i.e. T • 2nr, 
or T2rcr = jzTr + mg, 

or Tnr = mg. 

Fig. 67 

This is deduced, assuming static conditions to hold in the 
actual case. To a closer approximation Lord Rayleigh showed 

mg = 3- 8/T 

gives an expression of the relation between T and m. 

In practice the uncertainty may be avoided by taking the 

expression T = \f- and finding the value of K for the particular 

tube used. 

A tube is drawn out to a fine capillary of about *5 mm. 
at the end. It is connected to a burette by means of a 
short length of rubber tubing. The burette and tube are of 
course first made thoroughly clean, as in all the surface tension 
determinations, by the method given on page 42. A liquid of 
known surface tension, say water, is placed in the burette and 


the tap is opened to an extent which allows the drops to form 
at the rate of about one every second. A large number, say 
ioo or more, are collected in a weighed vessel and m per drop 
is obtained, hence, knowing T, K may be calculated. The same 
tube is now used for the liquid of unknown surface tension, say 
paraffin oil. Some of the liquid is run through the apparatus to 
remove all traces of water, and the burette is then filled with an 
uncontaminated sample *of oil. The collecting and weighing is 
repeated, and knowing K, T is calculated. 

Find by the above means the surface tension of alcohol, 
paraffin oil, benzine, etc. 

Fig. 68 

(3) Determination of the Value of the Surface Tension of a Solution 
in the Form of a Film 

The value of the surface tension can be obtained for, say, a 
soap solution film in the following way. 

Two pieces of copper wire, ABC and DEF, are bent as seen 
in figs. 68 and 69, so that AC = DF = about 4 or 5 cms. 
At the points A, C, D, and F a length of cotton thread is fastened, 
so that when the whole arrangement is suspended at B the 
thread takes the form of a rectangle about 5 cms. by 10 cms. 
(fig. 68). 

If now a film of soap solution be stretched across the thread, 
the latter will be pulled into the form shown in fig. 69, AD and 
CF taking the form of arcs of circles. 

Let AC = DF = 20 cms., GH = 26 cms., AD = CF = 2A 
cms., and the mass of the lower copper wire, DEF = m grammes. 

Let T be the value of the surface tension in dynes/cm. When 
stretched in this manner, there will be a tension in the thread 
equal to, say, / dynes. 

Suppose a is the angle which is included between the horizontal 
and the threads as shown in fig. 69. 

The film being two-sided, the vertical force upwards due to 
surface tension on the length 2a at the bottom of the film = 4«T. 



The resolved part of the tension / at D and F is / sin a upwards, 
i.e. total upward force of 2/ sin a. 

The downward forces are mg, the weight of the lower copper wire. 
We thus have for equilibrium : 

mg = 2/ sin a + 4«T (1) 

Considering the equilibrium of half of the film as obtained by 

a vertical dividing line through the mid-point of the film, we 
have the following equal and opposite forces acting : 

Fig. 70 

(a) A force '/ cos a ' at *^ e to P an ^ an e(ma * one at * ne bottom 

= 2/ cos a. 

(b) due to the two-sided film a surface tension effect 

= 4™, 
i.e. 2/ cos a = 4TA (2) 

From (1) and (2) we have tan a = 

_ mg — 4aT 




From the geometry of the case (see fig. 70), if O is the centre 
of the circle of which CHF is arc, a = z. OCIC 

So that tan a = ^§» 

and if HK — d = a — b, 

r -d 

tan a = 


Eliminating r since h 2 = (2r — d)d, 

h 2 + d 2 

r = 

h 2 -d 2 

We have tan a = ,, 


Equating this to the value given in (3) we get 

mg — 4aT ^ h 2 - d 2 

4T& ~~ 2dh 

2 d(mg) = 4T(A a - d 2 ) + 4*T • 2d, 

whence T = /T . ,f -, 

2(h 2 — d 2 + 2ad) 

Putting d — (a — b) 

Using the very simple apparatus described, a fair value of T 
maybe obtained simply from a knowledge of a, b, h, and m. 

A film is stretched across the string very easily by placing 
the frame horizontally in a flat dish containing the soap solution. 

Care must be taken to avoid excess of soap solution spreading 
over the part of the apparatus other than the thread. None 
must be allowed to remain on the lower copper wire, for obvious 

The measurement of the dimensions of the film can be made 
by means of ordinary dividers. The method does not justify 
the use of a travelling microscope. The measurements should 
be taken fairly rapidly so that they are all obtained before the 
condition of the film changes appreciably. 

In an example 2a = 3-5 cms., 26 = 1-2 cms., 2h = 11-7 cms., 
m = 1-51 grms., which gives T = 23 dynes/cm. 

(4) By Measurement of the Rise 0! a Liquid in a Capillary Tube 

If a clean, fine-bored capillary tube is depressed into a liquid 
Which ' wets ' it, and is then clamped vertically, the lower end 


of the tube being just below the surface of the liquid, it will be 
found that a column of the liquid remains in the tube, so that 
the surface in the latter is a height h cms. above the free surface 
of the liquid in the vessel which contains it. 

Suppose r is the radius of the tube and p the density of the 

The forces acting on the liquid in the tu£>e are : 

(1) The weight of the liquid. This is equal to (the volume of 
the liquid) X gp. 

Now the volume of the liquid is equal to V = nr 2 h + (volume 
of the meniscus) for a uniform tube. If r is small the meniscus 
is practically hemispherical, hence 

i.e. the downward force is 

V = *r*h-+ \{nr 2 ) r --t^ 3 ' 

3 j 


'(* + 0*p (5) 

(2) The Upward Surface Tension Force 

The line of contact is the intersection of the glass wall 
and the liquid surface, i.e. a circle of radius r. If a is the angle 
of contact the upward force, from the definition of surface 
tension, is equal to 

2-nr • T cos a (6) 

For equilibrium, the forces (5) and (6) are equal and opposite , 
2tztT cos a = itr 2 f h + - Jgp, 



T = gpr 


2 cos 

when the liquid wets the glass as in the case of water and clean 
glass a = o, 

^ T -££(»+£) (7) 

To find T, using a capillary tube, a glass tube is cleaned 
thoroughly. This may be done by using nitric acid (in which 
the tube is boiled) and caustic soda ; the tube is washed in 
tap water and dried in alcohol and ether, or the glass tube is 
allowed to stand for several hours, overnight if possible, in a 
concentrated solution of sulphuric acid (one part) and potassium 
bichromate (one part). It is then washed in tap water and 
dried. It 4s not advisable to use distilled water. 


The tube is then heated and drawn out to a capillary. A 
length of uniform bore is chosen and clamped vertically in a 
vessel which is brimful of the liquid whose surface is just above 
the top of the containing vessel, as shown in fig. 71. Care is 
taken to avoid touching the tube or the liquid in this adjustment, 
for even small traces of grease cause a large variation in the 
value of the surface tension. 

The tube is viewed by means of a travelling microscope, 
provided with a vertical traverse, the lower end of the meniscus 
is focussed and the vernier reading of the microscope noted. The 
free surface of the liquid in the containing vessel is next focussed. 
If the liquid surface is just above the top of the vessel, this 
level may be viewed very readily, and from the vernier reading 
on the microscope scale in this position the value of h may 
be obtained. 



Fig. 71 

To find r, the tube is broken at the point at which the meniscus 
rested, and viewed horizontally by the microscope. 

By arranging the cross-hairs »in the eyepiece to be tangential 
in turn to the two ends of a diameter, the internal radius may 
be measured on the vernier attached to the traverse. 

Alternatively, a weighed amount of mercury (Density, D =13-6) 
may be introduced into the tube and its length observed by 
means of the microscope, when the tube is horizontally on the 
bed of the microscope. 

Hence r, the mean radius of the tube, may be found if I is 
the length occupied by the mercury of mass m, for 

{nr 2 )lD = m. 

The experiment is repeated, using tubes of various diameter, 
and a mean value of T obtained. 

An alternative method of measuring h, which also overcomes 
the difficulty of viewing the liquid through the glass beaker, 
is one which makes use of a pin bent twice at right angles so 



that the point is displaced about one centimetre from its original 
position. The pin is attached to the capillary tube by means 
of a rubber band (see fig. 72). The point is in this way well 
removed from the curved surface of the liquid round the tube 
itself. It is adjusted to coincide with the free surface of the 
liquid. , . . . 

The image of the upper end of the pin is brought mto coinci- 
dence with the cross-hairs of a microscope, and the vernier reading 
on the vertical traverse is noted. The meniscus is next viewed, 
and the distance between it and the pinhead determined by the 
subtraction of the vernier readings. A subsequent measurement, 
by means of the microscope, of the vertical distance between 
the pin point and head enables the value of h to be obtained. 

Fig. 72 

(5) Surface Tension Determination from Measurements of Bubbles 

This method is suitable for measurements of surface tension 
of soap solution. 

Inside any curved film in equilibrium there is an excess of 
pressure over the outside, by an amount which depends on T, 
the surface tension, and R, the radius of curvature.* 

Consider a bubble with excess pressure p inside : take a 
section through the centre ; then there is an equivalent force 
over this section = p X (area) = ^>tcR 2 . 

The opposite surface tension forces, since there are two surfaces 
to the bubble, is 2T x (circumference of section) 

= 2T-2tuR. 

* It can be readily shown that inside a cylindrical./*/** (which has two surfaces) 
this pressure excess == -j^- 

If now the film has an equal curvature in the other direction at right angles— 

making spherical bubbles, there is a further pressure excess of ^, making a 

total ±£ = 

p the excess of pressure. 


For equilibrium these are equal and opposite, 
i.e. 4tcTR = ttR 2 £ 

P = 


A suitable form of apparatus with which to obtain a measure- 
ment of p and R is seen in fig. 73. 

This consists of a fairly wide U-tube, containing water, sealed 
to a T-piece, T. One arm of the T-piece is bent at right angles 
and terminates at B, where AB is parallel to the U-tube limbs. 
To the other arm at C is attached a piece of rubber tube, R, 
having a glass rod, P, which just fits it, and which can act as 
a piston. 

P c^* 
/ /^ T " A 




Fig. 73 



The piston P is withdrawn and the end B immersed in the 
soap solution ; then P is advanced slightly, causing a bubble 
to form at B. The excess of soap solution is drained off this 
bubble by touching any excess with a pencil, or the side of the^ 
dish which contains the solution. The rod is then further 
advanced, so that a bubble of, say, 1 to- 2 cms. diameter is 
blown. It will be noticed that the water in the U-tube takes 
up a position as in the diagram indicating a pressure inside the 
bubble in excess of the atmospheric pressure. 

This difference in level may be measured, using a microscope 
with horizontal and vertical traverse. 

Having read this, the microscope is moved until the bubble 
is focussed, first the image of one side, and then the other being 
brought in coincidence with the cross-hair. The difference in 
reading of the vernier of the horizontal traverse giving the dia- 
meter D, of the bubble, 

p = hg dynes per sq. cm., for a water manometer. 


Hence T = 




(6) Quincke's Method 

This method is most readily followed for the determination 
of the surface tension of mercury. A large flat drop is formed 
on a horizontal platform, and from its dimension, as measured 
by means of a travelling microscope, the value of T, the surface 
tension, and 0, the angle of contact of mercury on the platform 
may be calculated. 




Fig. 74 

We will assume that the radius of the flat drop, R, fig. 74, is 
large, so that the drop is truly flat, i.e. the pressure at a point 
just above and just below the upper surface is the same. 

Consider a section as in fig. 75, obtained by cutting it by a 
vertical plane passing through the centre, LMRN, and two 
parallel vertical planes, AINL and ESRM normal to LMRN. 
Let ACDE be the horizontal plane of maximum area in the 
drop, so that a tangent plane at AE is vertical. 







/^T "-- 


VV 'ff " " 

I N 

Fig. 75 

By considering the forces on the upper portion, AECDLM, 
which is in equilibrium, a value of T may be obtained in terms 
of the dimensions. 

The horizontal forces acting on this portion of the drop due 
to the remainder of the drop are 

(a) T-LM, left to right, 

(p) The hydrostatic pressure over the surface LMDC, from 
right to left varies from o at the upper surface to gph 1 

at CD, and is of a mean value " — , i.e. total force is ^— 

2 2 

X area LMDC, where p is the density of mercury, 

i.e. equating (a) and (p) 

T-LM = £- p - l • JMh\ 

T== ^i 2 





Considering now the equilibrium of the whole slab, we have 

in addition t© the corresponding terms, T • LM and ^--, where 


A is now the total height (fig. 74), a term T- IS • cos a= -T« LM cos 
(fig. 75), from left to right. 

or T-LM(i -cos«) =££^LM 


or, substituting the value of T from (8), 

. 6 h 

sin — = — -==— - (q) 

The experimental details necessary to carry out such a deter- 
mination are as follows : A clean glass slab, provided with three 
levelling screws, is first of all arranged horizontally, by means 
of a spirit-level. Mercury is placed on its surface in the form 
of a circular drop (fig. 76). 

-*- ^ 

Fig. 76 

N.B. — The mercury should be as free as possible from impuri- 
ties. To ensure this it should, if possible, be submitted to one 
of the methods of purification described on page 532. 

To obtain the values of h 1 and h, use is made of a travelling 
microscope, which is provided with a cross-hair. The cross-hairs 
are arranged so that one is vertical in the eyepiece ; this can be 
done by viewing a thin wire plumb line. The edge of the drop is 
then focussed. When the vertical cross-wire is tangential to 
the image of the side of the drop and the intersection of the 
cross-hairs is at the point of contact, such as A, figs. 74 and 76, 
the position on the vernier of the microscope is noted. Then 
by making use of the vertical movement the upper and lower 
surfaces of the drop can be focussed and the corresponding 
vernier reading will enable the values of h x and h to be ascertained. 


For liquids which wet the surface of the glass a simple modifi- 
cation enables the method to be utilised. 


A concave lens of over one metre radius of curvature is 
supported concave side downwards, on three legs inside a glass 
box, having one plane glass side. The box is filled with the 
liquid, and an air bubble is blown under the concave surface, 
using apparatus similar to that shown on page 141. A narrow 
bent tube is connected to R in place of the U-tube, etc. shown 
in fig. 79. The slight concave surface enables this air bubble 
to be blown without much trouble. The section of such a 
submerged bubble would be similar to that shown in fig. 74, 
when inverted. Then, if h 1 is the distance between the lower 
surface of the large air bubble and the plane AB, and if h is 
the total thickness of the bubble as measured from the lower 
surface to the plane of contact with the lens, all measured through 
the plane glass window by means of a microscope, equations 
(8) and (9) will give T and for the liquid in the box. 

(7) Rayleigh's Method 

This method depends on a measurement of the wave length 
of ripples formed on the surface of the liquid whose surface 
tension is to be determined. 

The velocity v of a harmonic disturbance on the surface of any 
liquid is given by* 

„= !A g+ ^I, (10) 

N2« 5 ^ Xp 

when x is the length of the wave, T the surface tension, p the 
density of the liquid, that is 

» = ^+£- T ) (I0 «> 

The surface tension is therefore seen to increase the effective 

4.7C 2 

value of g. For large waves the term ^--T can be neglected, 

x p 

and the velocity of propagation is <J-s. 

For waves of less than 1*5 cm. wave length, the value of the 
term involving T becomes more important, and when X is suffi- 
ciently small the velocity becomes more nearly equal to 



i.e. the first term in equation (10) becomes negligible cf. the 
second when X becomes very small. 
Equation (10) shows that when x = o, v = 00 , and when 

* See Poynting & Thomson's " Properties of Matter." 


X = oo , v = oo , between these values of x there is a minimum 
value of v, corresponding to a value of x, which is obtained when 

_X _ 2ttT 

2n S ~~ Xp ' 

X = 2n-J— ; (n) 

this mmimum velocity is therefore 



For water x OT = 17 cms. from equation (11), for T =75, p=^i, 
g = 981, and hence 

v m = 23 cms. per sec. 

Waves having a smaller wave length than this critical value 
which corresponds to the minimum velocity are called ripples, 
and the more important term is the one involving T. 

Such ripples cannot be viewed and X measured directly, but 
the following method, first used by Lord Rayleigh, may be 
employed with success. 

The liquid is placed in a large flat dish, say, a large porcelain 
developing dish. As in all surface tension experiments, the dish 
is cleaned thoroughly before introducing the liquid. 

The ripples are made by having a tuning fork electrically 
maintained, arranged at one end of the dish as in fig. 77. Attached 
to one prong of the fork and dipping in the liquid is a thin 
sheet of aluminium foil, D. If the frequency of the fork is 
n, and X the wave length of the ripples we have 

v = n\. 
But the ripples cannot be directly observed and measured, as the 
phase change is rapid, and a general illumination results. 

If these ripples are viewed by intermittent light, the frequency 
of the flashes of illumination being the same as the source of 
the ripples, then they appear stationary and may be measured. 
(Cf. stroboscope, p. 404.) Alternatively, if the observer has 
intermittent views of the surface of the liquid, such views being 
of same number per second as the vibration number of the 
tuning fork which causes the ripples, between each view the 
ripples will have moved forward a distance equal to a wave- 
length, and if illuminated by a constant source of light they 
will appear stationary. 

To obtain such intermittent glimpses of the surface, the latter 
is observed through the prongs of a second tuning fork of the same 
frequency, and maintained in vibration by the same circuit. 
Two thin pieces of aluminium foil, A and B, are attached to the 



prongs of the second fork, so that direct vision is impossible when 
the fork is at rest, but is obtained when the prongs of the fork 
are at the position of extreme separation. Thus, for each 
complete vibration of the fork a view is obtained of the surface, 
and an apparently stationary train of waves is seen when viewed 
in this way. The wave length can be obtained by direct measure- 
ment of the longest possible number, m say, by dividers 
adjusted over the surface. 

When performing this experiment with water, the measurement 
will be found to be somewhat complicated due to shadows cast 
by the water, and the exact setting of the dividers over the 
surface for the stretch of m waves will not be as easy as in the 
case of a more opaque liquid. 

Fig. 77 

A method which has been found to give satisfactory results 
is to suspend an incandescent bulb about two metres above the 
surface of the water. This casts a series of shadows on the 
bottom of the white porcelain dish. The dividers are arranged 
also to cast a shadow and adjusted so that the two ends of the 
dividers' shadow coincides with the corresponding parts of the 
first and last of the m waves. In such a case the magnification 
of the shadow of the waves due to the obliquities of the incident 
beam is compensated by a corresponding distortion of the shadow 
of the ripples, and the kind of thing shown in fig. 78 results : 
the length of, say, ten or more ripples may be obtained, whence x 
may be calculated. 

Now, if n is the frequency of the fork, 


T=^(2M a X*-£). 


A suitable experiment by such a method is to find the variation 
of surface tension of a salt solution with concentration. 

Use first of all pure water. Make sure that the aluminium 
plate is clean, and that no soft wax, by which it is fastened to 
the tuning fork, adheres to it, and so contaminates the water. 

Measure the value of X and calculate T, n being known. 

A suitable frequency for the two tuning forks is about 60. 

Then repeat the experiment with a sodium chloride solution 
having \, 1, ij, to 5 grm. molecules per litre, and plot a graph 
showing the increase in the value of T with concentration of 
the solution. 

Fig. 78 
(8) Jaeger's Method 

In this method air bubbles are formed in the liquid under 
investigation. The bubbles in this case have one surface only, 
and therefore the excess of the internal pressure over the external 
assuming a hemispherical form is 

* 2T 

The value of R is fixed, and equal to the radius of the orifice 
from which the bubbles emerge (approximately). 

The following apparatus, as seen in fig. 79, is used. 

A thin glass tube, C, drawn out to a fine capillary of less than 
*5 mm. ('2 to *5 mm. is satisfactory), and is fixed vertically in 
the liquid. It is connected either rigidly, as in the diagram, 
or by means of a piece of india-rubber tubing, to glass tube, AB, 
which terminates in another rubber tube, R, which is clamped 
in a screw pinch cock, T. 

Sealed into AB is a glass T-joint which leads to a manometer U. 

Air is sent in at D at a steady rate, and consequently, bubbles 
are formed in the liquid. To ensure that they are always formed 
at the same depth, h cms., below the surface of the liquid, a 
scratch is made on C, and this is adjusted to coincidence with 
the liquid surface. 

When the bubble is of the same diametei as the orifice at G, 



it becomes unstable and breaks away. During the formation 
of the bubble the manometer rises and the maximum difference 
in heights between the two columns, H cms. , is observed for many 
bubbles, the pressure of the incoming air being constant ■ * 

When a good agreement for H for a series of bubbles is obtained, 
the tube C is taken out and the value of R determined by 
viewing the end with a microscope. If the capillary is very 
small, a microscope carrying a $mall scale in the eyepiece is 
used, and the value of the scale divisions in the eyepiece scale 
determined by comparing with the image of a small mm. scale 
viewed by the microscope. 











B i 

Fig. 79 

The tube is rotated and other values of the ' diameter * 
taken. These measurements should be taken before commenc- 
ing the experiment to ensure a regular tube. 

A good method of producing the stream of air is to allow 
water to enter, one drop at several seconds interval, into a 
Winchester, closed at the top by a cork, which is waxed into 
the neck of the bottle and provided with an entrance and exit 
tube. The types shown in figs. 79 and 80 serve very well : the 
air is trapped inside the bottle by water in the bottom of the 
bottle in the form shown in fig. 79, and in the U-tube entrance 
in the type illustrated in fig. 80. 

In this way the pressure gradually rises until of sufficient 
magnitude to form a bubble of maximum diameter. This 
detaches and an interval then occurs, during which the pressure 
again rises, as is seen in the manometer U, and the whole process 
is repeated. 

By slowly increasing the pressure, its maximum value corres- 
ponding to H cms. difference in the manometer may be easily 

*It -might be. found advisable to draw out the open end of the manometer U, 
to a capillary tube, to damp the oscillation of the liquid. 


measured either directly on a scale or by means of a travelling 

If only one large bottle is available, air is compressed into it 
and the outflow tube of rubber connected to B. By varying 
the compression on this tube by means of a screw clip the condi- 
tions described above may be obtained. 

Let H be the maximum pressure difference established in the 
manometer, and h the depth at which the orifice is placed below the 
free surface of the liquid, then the pressure of the air inside the 
^bubble, when the latter is about to leave the orifice, is n + gaH, 
where n is the atmospheric pressure and a the density of the 
liquid in the manometer. 






Fig. 80 

The pressure outside the surface of the bubble is n+ gph, 
where p is the density of the liquid under investigation. Then, 
P being the excess of pressure inside the bubble over that on 
the outside 


p = (n -f-goH) - (n + g?h) = -^ , 

or* T = ?£(aH-p/*) (12) 

The values of R, H, and h, are obtained as already described, 
and T is calculated. 

The method is a good one for obtaining relative values of 
surface tension. As an example, find the variation of surface 
tension with concentration for a salt solution containing from 
o to 5 gramme molecules per litre. 

* The result expressed in equation (12) above is developed assuming the bubble 
formed is hemispherical and of the radius of the tube. 

Ferguson ("Phil. Mag.," No. 28, 1914, page 128 et seq.) has deduced an 
expression for T without making this assumption and arrives at the result : 


I »Vb/ 



By altering the temperature of, say, water in the beaker, the 
change of surface tension with temperature could be obtained. 

(9) Capillary Tube Method (Sentis). 

A capillary tube is drawn out to about *5 mm. bore as in 
Jaeger's method. It is immersed in the liquid under investiga- 
tion, and then withdrawn and clamped vertically. Some of the 
liquid will emerge at the lower end and form a drop as shown 
in fig. 81 (1), so that the distance from A, the meniscus in the 
tube, to B, the lowest point of the drop, is h x cms., and MN is zr. 

If now the lower end of the tube is surrounded by a vessel, C, 
containing the liquid, the column will fall in general, but the 
meniscus may be brought to the original level by raising C until 
the free surface of the liquid in the beaker is h 2 cms. below the 
meniscus, fig. 81 (2). 

Fig. 81 

From a knowledge of h x , h 2 , r, and p, the density of the liquid, 
T, may be Calculated from the formula 

T=f{r(/> 1 -^)-^.. 

To establish this formula we may assume that the portion 
of the drop, shown in section as MONB, is hemispherical. This 
approximation is a safe one when the radius of the capillary tube 
is small, and r small compared with h v 

Consider the forces acting below the horizontal plane of 
maximum area shown in section as MN. 

The length OB = r, and hence the column from MN to the 
meniscus is (h t — r) ; of this a length, h 2 , is supported by the 
upper surface tension forces as shown in fig. 81 (2), and hence 
the pressure at MN due to the liquid column is g?{h 1 — h 2 — r), 
which contributes a downward force g9{h 1 — h 2 — r)nr 2 . The 

weight of the hemisphere - T:r 3 gp also acts downwards. 


The surface tension acts vertically in the circle of section of 
the drop and plane MN, and has a value T • 2izr ; hence 

T27tr = gpfa — h t — r)*r* + -*r 3 gp, 

To make a determination of T for a liquid, the cleaned capillary 
about 25 cms. long is drawn out to a radius of about «5 mm. 
at the one end, and is almost entirely submerged in the liquid, 
so that the latter fills the greater part of the bore. It is slowly 
withdrawn and is clamped vertically. By means of a travelling 
microscope, the length MN = 2t is measured and the microscope 
is then focussed on the meniscus. A small glass beaker is 
raised under the tube, until the liquid in the beaker just touches 
the drop. The reading of the micrometer screw, which raises 
the platform carrying the beaker, is noted. The beaker is then 
further raised until the meniscus, as viewed by the fixed micro- 
scope, again acquires its original level. The micrometer screw 
reading is again noted. The difference between these two 
readings is equal to (h x — h t ). 

P is determined in the usual manner, whence T may be calcu- 
lated from the formula deduced above. 

If the form of adjustable table with micrometer or vernier 
attachment is not available, some simple convenient method 
may be devised for the measurement of (h x — h t ). For example, 
two microscopes may be used. With the first the value of MN 
is observed, and then the point, B, is viewed and its image 
brought into coincidence with the cross-hairs; the second 
microscope is adjusted until, viewing the tube conveniently at 
right angles to the first, the image of the meniscus is in coincidence 
with the cross-hairs in the eyepiece. 

The beaker, C, is then introduced and adjusted until the 
meniscus is again as before, producing an image in coincidence 
with the cross-hairs of the second stationary microscope. A pin 
is adjusted to coincidence with the free surface of the liquid in 
the beaker which is then removed. The first microscope is then 
moved a distance which is measured on the vernier scale attach- 
ment, until an image of the point of the pin is in coincidence 
with the cross-hairs, the vertical distance moved by the microscope 
is {h x — h 2 ). 

Care is, of course, taken that the capillary tube does not move 
during the experiment. 



(10) Anderson and Bowen's Method i 

A method by which to determine the value of the surface 
tension of a liquid, and the angle of Contact with glass was 
described in the " Philosophical Magazine," April, 1916. Another 
method not so readily adaptable to general laboratory use is 
seen in the same magazine, February, 1916. 

A small rectangular sheet of thin cover glass is cleaned (by 
standing it in concentrated sulphuric acid and potassium bichro- 
mate, etc. as described previously) and dipped into the liquid 
whose surface tension is to be measured. It is withdrawn and 
clamped vertically. The liquid takes up the form shown in the 
diagram, fig. 82.* 





O l 


Fig. 82 

The drop has two curvatures, making the equivalent of a 
cylindrical lens, concave at the upper half and convex below, 
the centres being at O and O 1 . 

The upper limit of the drop may be at O or any point, N, 
above. The drop is tangential to the glass plate at O. 

If A is the focal point of the concave lens, OA the axis, f x the 
focal length, B the focal point of the convex lens, and OB the 
axis of the lens of focal length, / 2 , 

OO 1 = 

r, = 







h cms. 

the radius of curvature of the concave 
(assumed symmetrical and equal), 
— the radius of curvature of the convex surfaces, 
= the refractive index of the liquid, 
= the density of the liquid, 
= the pressure in the liquid at O, 
= the pressure in the liquid at O 1 , 
= atmospheric pressure, 
= surface tension of the liquid. 

* It was established by the original experiment that for water, glycerine, 
olive oil, and turpentine, the angle of contact is zero, and hence the form of 
fig. 82 represents the section of the drop. 

We have, as the lens is a thin one, using the lens formula 
/■>(!*- *) (7 ~ 7). (page 278.) 

7i = { *- x) k (I3) 

rr { *- J) k --m 

Since the pressure inside a cylindrical surface is greater than the 

pressure outside by p = T Q-), (see Poynting and Thomson's 

" Properties of Matter"), and here we have a cylinder of radius r, 
i.e. R x = r 

^ = U -J- 1 (i5) 

^ = n+l .. (l6 ) 

Now p2-pi= gph, 

Pt ~ Pi = T ( - 4- - J by (15) and (16) above. 

Hence t(J- + ±-J = g9 h. 

But by (13) and (14) above 

1 _ 1 1 _ 1 

*1 2/lOx-l) ' y2~2/ 2 ((X -I) 1 




^)(7x + 7 2 ) = ^ 


or T - 2gp ^^ ~ I ^ 1 ^ 2 

(/i+/ 2 ) 

If one side only of the glass sheet is wet, using the same notation 
we have 

T _ gP%-i)A/ 2 

(/!+/■) (l8) 

The apparatus used to obtain T, is a collimator illuminated 
by a sodium flame, and adjusted to give a parallel beam. The 
light passes through the cover glass and the liquid lens, and is 
viewed by a low-power microscope provided with a vertical 
traverse, and a traverse parallel to the axis of the microscope. 

The usual type of travelling microsope will need a littie 
modification to make this latter condition possible. 

If a parallel beam of light be sent from the collimator from 


left to right, and normal to the plate, a virtual image of the 
horizontal slit will be formed at A by the upper half of the 
liquid lens. The distance OA may be measured by using the 
low-power microscope, arranged with its axis parallel to the 
direction of the incident beam. The microscope is first focussed 
on the image at A, and then moved backwards a measured 
distance until the glass plate is in focus. The distance moved 
being = OA = f v In the same way OB and/ 2 may be measured 
by noting the difference in reading of the microscope when the 
glass sheet is focussed and then when the image of the slit is 
coincident with the cross-hairs. 

It will be found most satisfactory to use one side of the cover 
glass only, i.e., dry the other side before making the observation, 
and allow the incident beam to fall on the dry side. 

As the incident beam is parallel, it will be found, of course, 
that the observing microscope must be moved in a vertical x 
direction to enable a focus of first A and then B to be made. 
This distance, h, is noted. 

The refractive index and the density may be obtained from 
tables, or by one of the many methods available. 

Thus, having measured / x , / 2 , and h, knowing jx and p, T may 
be calculated for the liquid used. 

Measure in this manner the surface tension of water and 

The Variation of the Surface Tension of a Liquid with Temperature 

The variation of surface tension with temperature may be 
obtained by Jaeger's method, which enables a good comparison 
of the relative values of the surface tension at different tempera- 
tures to be made. 

The details of the experiment are as described on page 140. 
The bubbles in this case are formed in the liquid at different 
steady temperatures, and H is determined for each. 

A large beaker is rilled with the liquid, say water, and heated 
to about 90°C, and then allowed to cool. The value of T being 
obtained every io°C. The liquid is well stirred before each 
observation, and if a large volume is taken will remain sensibly 
at the same temperature throughout the observation. 

A curve is plotted, showing the decrease of T with temperature. 

Capillary Tube Method 

The decrease of the value of the surface tension may also be 
investigated by a capillary tube method. Either form of appara- 
tus shown in fig. 83 or fig. 84 may be employed. One 
form of apparatus is filled with the liquid and immersed in a 
water bath whose temperature can be regulated either by a 


thermostat or by manipulation of the Bunsen burner and a 
stirrer. The bath is raised to boiling point and allowed to cool, 
so that at about 90-95 °C. the whole of the apparatus and contents 
are at the temperature of the bath. The difference in level 


Fig. 83 


between the two surfaces, A and B, is measured as quickly as 
possible with a travelling microscope. This process is repeated 
at different temperatures, say, every io°C, and from a knowledge 
of this difference in level, h and the radius of the tube as measured 


Fig. 84 

by the method given on page 132, the value of T at each tempera- 
ture may be calculated and a graph representing this relation 

The value of the density (p) is obtained, for each temperature, 
experimentally, or from tables, and is used in the evaluation 
of T by the above methods. 

The essential to success in this and every other surface tension 
experiment is that the glass, etc., is clean. 



WpEN adjacent layers of a fluid move with a relative velocity, 
forces, known as viscous forces, are brought into play tending 
to reduce this relative movement. 

If we consider a fluid whose upper layer is moving with a 
velocity v in a fixed direction, the state of affairs shown in fig. 85 
will be reached, where intermediate layers, between the upper 
layer AB, which has a velocity v, and the lower layer CD, which 
is at rest, have a velocity shown by the arrows. 

L*ve» .in N/v©-no«N» 

Fig. 85 

The force, F, acting on any area in a plane at right angles 

to the diagram, and parallel to EF, is proportional to the area A ; 

and to the velocity gradient, in the case taken -y 1 i.e. at constant 


F oc A x (vel. gradient). 
Taking the normal to EF, in the plane of the diagram, as a y 
axis we have 

F-'.A*. W 

where t\ is a constant for the liquid and is called the coefficient 
of viscosity. 

In the case of a liquid flowing down a tube, the axial stream 
is moving with a definite velocity and the layers in contact with 
the walls of the tube are at rest and, provided that the pressure 
difference which is causing the flow is not too great, the result is 
the regular type of motion already considered. 

If the pressure exceeds a certain limit, the liquid no longer 
proceeds in this regular manner, i.e. no definite stream-line flow 
takes place. The result in this case is called turbulent motion. 



We will assume, in the experiments that follow, that the 
pressure applied is below this critical pressure, and that the 
motion is therefore regular. 

The Determination of the Coefficient of Viseosity for a Liquid by Obser- 
vation of the Flow of the Liquid through a Tube 

The value of the coefficient of viscosity of a liquid, such as 
water, may be obtained by measuring the quantity passing per 
second x through a tube of uniform radius, when a definite 
pressure difference exists between the ends of the tube. 

Consider, as in fig. 86, a section of such a tube, whose radius 
is R cms., and imagine a thin tube of liquid in it of radius, r, 
and thickness dr ; the area of cross-section of such a tube is 2nr -dr. 

Fig. 86 

If P be the pressure difference between the ends, then the 
force actmg on the tube ABCD, due to this pressure, is P • 2nr • dr. 

Over the curved area of such a hollow cylinder there are 
viscous forces which are of a magnitude dependent on the value 
of the distance r from the axis of the tube. 

Fig. 87 

Fig. 87 shows the form of distribution of velocity of the layers 
for various values of r ; as shown later, equation (3), the curve 

of fig. 87 is a parabola, and hence j- is proportional to r, and 

therefore the viscous force, F, is also proportional to r by (1). 

dr ex P ress the rate of cn ange of this force with r, then 


the value of the difference between the magnitudes at the two 

surfaces of the thin tube ABCD, since the thickness is dr, is j- • dr. 

In steady flow we have, therefore, the equal and opposite 
forces, which may be expressed as : 


2izrP - dr, 

-^=27^ (2) 

Now equation (1) gives 

The area A is the area over which F acts, i.e. is 2itr • I, I being 
the length of the tube. Equation (2) becomes 

r [ -117 2tzH J = 27crP. 
it V dr J 



,dv r % „ , n 

— 2l\Ttrl -j = 27C — • P + C, 

dr 2 

C being the constant of integration. 

But when r = o, -j — .*. C = o, 

i.e. 7 dv _, 

— 2vjZ 3- = rP. 

Integrating once more 

— j 2t\1dv = j r¥dr, 

— 2lf\V = — -f B, 

PR 2 

When r = R, v = o /. B = 

Hence » = -— (R 2 - r 2 ) (3) 

In one second, a column of liquid, v cms. long, and 2-nr • 87 cross- 
section flows, down such a hoUow tube, i.e. the volume passing 

per second = j-y- (R 2 — r 2 ) I 2itr • dr. 

Thus, for the whole tube of radius R, we have the sum of 
such expressions. 


If Q is the total amount of liquid passing per second 

/.r p 
27c — (RV - r*) 8r 






4 J< 

W 8 I r> 


A suitable form of apparatus to use in an experimental deter- 
mination of T) by this method is seen in fig. 88. The liquid, say 
water, is contained in a large bottle, B, standing a suitable 
distance above the level of the table. The water flows from 

Fig. 88 

this reservoir to the union, X, thence through a capillary tube 
of known length to the union, Y, and so on, via a length of 
india-rubber tubing to a graduated jar, J, where the thermometer, 
C, measures the temperature of the emerging water. 

From the unions, X and Y, two lengths of india-rubber tube 
make connection to the manometer M. The difference in the levels, 
E and F gives in cms. of water the value of the pressure difference 
between the ends of the experimental tube, K. 

A pinch-cock L enables the flow of the liquid to be regulated. 

In order to maintain a constant difference of pressure between 
the two ends, X and Y, whilst the water is flowing, the bottle B 
is closed by means of a tight-fitting india-rubber cork through 
which a glass tube passes, to a point well below the surface of 
the water. 'This end being open to the atmosphere allows the 
entrance of air bubbles as water flows through the tube. The 
lower end of the tube remains at atmospheric pressure, and so, 


until the whole of the water above this point has passed through 
the tube, the manometer will remain with a uniform average 
difference of level. 

The flow should be so arranged that the emergent water 
issues as a slow trickle or succession of drops, avoiding a rapid 
stream of water, which might cause the flow in the tube to become 
turbulent, in which case the formula which is developed, assum- 
ing a regular flow, will break down. 

In adjusting the apparatus, the water passing is not collected. 
When everything is steady the tube is inserted into J, as seen 
in diagram, and the time is taken in seconds for a definite amount 
of water, say 500 c. cms., to pass. From this, Q is obtained. 

If h is the difference in level in the water manometer, P = g?h, 
approximately, or as shown later, equation (n), 

more exactly. 

Before determining the remaining unknowns, I and R, several 
values of Q, corresponding to different values of P, should first 
be obtained, for we have 

_kR^ P 
^"8 l"Q' 

By varying P and Q, the mean value of -~- can be obtained, 

and this, not one value of P and corresponding Q, used for the 
computation of 73, provided that the temperature of the room remains 
constant during the experiment. 

The value of / may be obtained by direct measurement of K. 

Now R, wjiich occurs in the formula in the fourth power, 
must be obtained as accurately as possible. A suitable method 
is the measurement of the length and mass of a column of mercury 
in the dried tube as described on page 132. 

Unless the liquid is passing through the tube with a very 
small velocity, equation (4) must be modified to allow for the 
kinetic energy imparted to the liquid. This reduces the effective 
pressure P. 

To arrive at such a correction, consider the case of a liquid 
flowing through an irregular tube whose two ends are at different 

Take two points, B and C, in the tube such that the cross- 
sections are A t and A 2 ; the velocity of the centre of the liquid, 
v x and v 9 ; the pressure in the liquid, P x and P 2 ; and the dis- 
tances of the centre at these places are L x and L 2 cms. below a 
fixed horizontal plane. 


If the liquid moves a distance dx x at B, it will advance a dis- 
tance dx 2 at C, such that 

A 1 dx 1 = A 2 dx 2 (5) 

The work done on the liquid between B and C is therefore 

PiAji*! — "P 2 A 2 dx 2 (6) 

This work must be the equivalent of the energy gained in 
the liquid. 

Gain of Kinetic Energy 

The mass (A^xjp of liquid enters at B (p being the density 
of the liquid, assumed constant). This has a K.E., 

I A x dx x ^v x 2 . 

Similarly for the mass of liquid leaving, and since condition (5) 
holds, the g ain of kinetic energy is 

Aiteiftpw," - IpvS) (7) 

Fig. 89 

Potential Energy 

The P.E. of the mass of liquid at B is gL 1 (A 1 ^ 1 )p, and at 
C is gLgAjjtaaP. 

For a gain of P.E., C must be higher than B, L 2 <Lj,i.e. 
the gain of P.E. is 

A 1 dx 1 (pgL 1 - P gL 2 ) (8) 

Equating (6) to the sum of (7) and (8) we have, using (5) 

A x dx 1 (P 1 - P 2 ) = A l dx 1 {$ 9 v 2 * - i pVl *) + A 1 ^( P ^L 1 -p^L 2 ), 
re-arranging terms, 

Pi + *P» x 2 ~ gtU = P 2 + |pV - £pL 2 , 

i-e. P + f p*> 2 — g P L = constant (9) 

Applying to the case shown in fig. 90, where the liquid is 
contained in a wide tank, BC, and flows .through a uniform 
tube, CD, with a constant maximum (central) velocity, v, we 
may find the effective pressure difference between the two ends 
of the tube using equation (9) above. 



If II is the atmospheric pressure we have, applying (9) to the 
point B and C, where C is just inside the tube and is at pressure p', 
H + ipv v i-gpL B = p' + l ? v* - £ P L C 

For such a wide cross-section as the tank we may assume the 
surface to be fixed. This may be further brought about by the 
device shown in fig. 88. 
i.e. v B = o. 

We will further use this surface as reference plane for measure- 
ments of L, i.e. L B = 0, L c = L cms. 

We therefore have 

II =P' +$ p v*-g P L, 
or p' =11 +g p L-|p*;2, 
the pressure at the open end, D, being n, we have the pressure 
difference between the ends of the tube, CD, p, = p* ~n 
i- e - P =£pL-| P t/ 2 ( I0 ) 

Thus, the effective ' head ' is reduced due to the gain of kinetic 
energy by the liquid. 


C D 

Fig. 90 

It was seen (p. 151) that the velocity of the stream at a distance 
r from the axis of the tube, was 

2ril\ 2 J' 

i.e. when r = o, we obtain the value of v used in equation (10) 
above, i.e. the maximum velocity, 

PR 2 


Now we saw (p. 152), 


P 7C R* 



rR2 ' 

and the effective pressure, p, is given below, 

*=* p ( L -i^r.) («) 


An alternative apparatus for the determination of tj is seen 
in fig. 91. The horizontal capillary, DE, is fixed in a cork 
which closes the lower end of the tube, CF, which has con- 
strictions at A and B, on which are scratches on the glass. The 
whole vessel is filled with the liquid and the ground-glass stopper, 
S, placed in the neck above A. The volume between the scratches 
is calibrated, so that its capacity, V, is known. The stopper, S, 
is removed, and the time, t cms., is taken for this volume to flow 

Fig. 91 

through the tube, i.e. when the level reaches A a stop-clock is 
started, and when the level reaches B, the clock is stopped. 

p, the pressure, is taken as due to the average height, h, of 
the liquid above the capillary tube level, i.e. p = gph, or as 
shown on more precisely in equation /n), 

The other terms are measured as for the first form of apparatus 

Determination of the Viscosity of a Liquid by the Coaxial Cylinder 

The value of the coefficient of viscosity for a liquid such as 
glycerine may be obtained, using the apparatus shown diagram- 
matically in fig. 92. The liquid, say glycerine, is placed in the 
cylinder, AB, which may be rotated by hand or by a small motor. 
A belt driven by either means passes over the pulley, P, and the 



rotation is imparted to the cylinder by the crown bevel wheels, 
W. The revolutions may be counted by the revolution counter, R. 
Hence, if a number of revolutions, n, be timed, the angular 
velocity, w, may be calculated. 

Immersed in the glycerine is a second solid cylinder, CD, 
which is suspended on a phosphor-bronze suspension, which 
carries a mirror, M, and which is maintained central by the pivot S. 
Due to the viscous forces in the liquid, the inner cylinder will 
experience a couple, C, which turns it through an angle, 0, such 
that the torsional restoring couple just balances the turning 
moment due to the liquid. 

If t is the restoring couple per unit angular twist in the sup- 

porting wire, and 6 be the constant deflection in radians, we 
have t0 = the restoring couple due to the torsion of the wire. 

To obtain a value for the couple due to the viscous forces, we 
will first consider the co-axial cylindrical surfaces. 

Let fig. 93 represent a section normal to the axis of these 
cylinders, of radius R 2 and R, as shown. 

Take any thin cylindrical ring in the liquid having points, 
A and B, on the same radius. If the liquid rotated as a whole, 
there would be no velocity gradient — no relative motion. Thus, 
for no relative motion, when A moves to A 1 , B moves to C. If 
w is the constant angular velocity, AO = r, and BO = r + Br ;, 
AA 1 = rw ; BC = (r + 8r)w. Now, in the actual case, 
the outer cylinder moves with an angular velocity, Cl say„ 
and the layer in contact with the inner cylinder is at 


rest, i.e. the liquid does not have this constant angular velocity. 
Actually, the particle at B moves with some larger angular 
velocity {w + dw), and B moves to B 1 where BB 1 = (w + dw) 
(r + dr), thus having an excess over the no-relative motion 
velocity equal to (w + dw) (r + dr) — w(r + dr), i.e. = 6w(r+dr) 

i.e. the velocity gradient is — ^~ — '-, or since dr is negligible 

cf . r, the velocity gradient ~ = r • — in the limit. 

Thus, considering the forces on the cylindrical shell, if / is the 
length of the inner cylinder from the surface of the liquid to 

Fig. 93 

the lower extremity, the area of the cylindrical shell is (2^)/, 
and therefore from (1), 

F = yi(2*rl)£> 

where F is force due to the viscosity over the curved surface 
of such an imaginary shell. The moment of this about the common 
axis is 

C = Fr — 2itrHt\r ^—» 

substituting the value for -j-. 

Integrating we have 

J2nlridw =J~-JT> 

i.e. 2ir/i)tt> =* - — % + B, 

where B is a constant. 
Now, when r = R lt w = o, 


r = R,, w = H, 2R X * 


i.e. 2 ^n=§[^-^ 5 ], 

Also we have 

L== R^-R** (I2) 

C = T0, 

Wh6nCe " - ^OR.'R. 1 ' " - (I3) 

In the actual case taken (12) must be modified to include the 
couple due to the viscous effect on the under surface of the 
inner cylinder. If this effect were truly due to the parallel 
circular plates, we might readily deduce an expression for it. 
But the stream-lines in the liquid will not be composed of two 
such regular patterns. The effect at the end of the cylinder, 
CD, will be a gradual transition from the one to the other. 
Some end correction is therefore needed. Let us assume that 
the couple exerted on the cylinder, CD, due to the under surf ace 
and the lower portion of the curved surface, and also the "effect 
due to the support, S, which rests on AB, for an angular velocity, 
CI, is K CI, then the total couple will be 

_ WRi 2 R a 2 Ka 
C ""(R,»-R 1 *)- Q + K0 ' 

where V is the length of the curved surface of the inner cylinder, 
measured from the liquid surface to the lower limit of the regular 
stream-line between the cylinders ; the second term deals with 
the rest of the curved surface and the lower surface, etc., as 

If now we have the space between the Cylinders filled to a 
level, such as L 1} we have, if l x ' is the corresponding length and 
B x the deflection 

^=°mw+«i- <-> 

For a second case, where the length of the cylinder immersed 
is Z 2 ', corresponding to a level, L 2 , in the figure, if 2 is the 

*- pg£g! + K] ( I5 ) 

K may therefore be eliminated from (14) and (15) and, if the 
angular velocity, Cl, is the same for both determinations, we 
have by subtraction 



i- e *>- fi R 2 2 -Rx 2 L i 2 J ( } 


The length to be measured is therefore the difference between 
the two levels, Lj and L 2 . 

Thus, by maintaining the speed of rotation constant, and 
observing the steady deflection corresponding to two levels of 
the liquid surfaces, knowing L^ = // - /,', and the dimensions 
of the cylinders, tj may be calculated in terms of t, which itself 
may be evaluated by observation of the twist of the wire, when 
the liquid is not in the cylinders, and loads are applied at G, as 
indicated in the diagram. 

For liquids whose viscosity is fairly large, the torsion control 
will not be sufficient. In this case the restoring couple is in- 
creased by adding masses, m, m, thereby increasing the restoring 
couple by m D, where D is the diameter of the wheel G. 

In many cases it is advantageous to eliminate the torsional 
restoring couple entirely. This may obviously be done by 
adding masses, m, m, of such magnitude that the inner cylinder 
is brought to its rest position, as observed by the reflected beam 
from the mirror, M, on the scale, which was previously used to 
measure B x and 2 . 

In that case, if masses of total value, M x , are supported on 

the wheel, G, C t = Mj • — , and for the second case with the liquid 
at level, L 2 , C 2 = M 2 -. 
Then equation (16) simplifies to 

(M, - M a )| - "gp%^ (// - V). 

This latter method of working will be found most satisfactory 
for glycerine and similar liquids, whilst the former method, using 
the torsion of the fibre, will be satisfactory in the case of less 
viscous substances, such as water. 

Viscosity of a Liquid (Oscillating Disc Method) 

The viscosity of a liquid may be determined by timing the 
period of oscillation of a flat circular disc in air, and finding 
the logarithmic decrement in the liquid and in air. 

O. E. Meyer has shown that with such a disc, the coefficient 
of viscosity for a liquid, tq, is 

7u P T(y 4 + 2r 3 ^) 2 \\ 7c °j + V 7c °j j " * ' * (I7) 

where I is the moment of inertia of the disc, and attachments 
about the axis of suspension, 
p the density of the liquid, 
T the time of a complete swing in air, 

■n = 


r the radius of the disc, 

d the thickness of the disc, 

x*the logarithmic decrement in the liquid, 

x the logarithmic decrement in air. 
The development of the above formula is beyond the scope 
of this book, and may be found in " Poggendorf Annalen," 
No. 113, page 55. 

Without any discussion of the development of the result, we 
will use it as an empirical formula which agrees with determina- 
tions by other methods. It is an excellent method whereby to 
study the determination of the logarithmic decrement of an 
oscillating system. 

Logarithmic Decrement 

Consider a suspended body to oscillate about the suspension 
as axis in a simple harmonic manner. If I is the moment of 
inertia of the body about "the axis of suspension, F the restoring 
force per unit angular displacement, we have the equation of 

IB -f F0 = o. 

If now a frictional resistance acts on the body so that K is 
the resulting opposing couple per unit angular velocity, the 
above equation is modified into the form 
I'd + Kb + F0 = o. 

By the method given, p. 28, trying as solution = djt, 

we arrive at the result, 

-J£« / /F K 2 \ 

e=6 .e *i cos^yjy-— t + aj, 

when O and a are arbitrary constants, being the angular 
displacement at any time t. 
The period time, T, of such a motion is 

T 2" 

I F K* 
\ I 4I 2 


The amplitude is o e al * , and decreases exponentially with 
the time. O is the amplitude when the friction is eliminated, 
i.e. when K = o. 
In the case where the friction resistance, K, is very small, 

K 2 
the value of K 2 is correspondingly much smaller and -p- becomes 

negligible compared with j . 

The periodic time becomes 

The oscillating body therefore performs vibrations in equal 
times, but the amplitude gradually dies out (see p. 29). 

Starting from the undisturbed position of the body, let a x 
be the angular displacement measured to the first turning point 
(measured to the right, say). 

Let a 2 be the angular displacement following, on the other 
side of the zero (to the left, say), also measured from the zero 

a 3 the next swing in the original direction measured from 

the zero. 

• a/ of+fit* +ir«/i *"* _ • 

The first deflection 0^ is after a time t = — ; <x 2 after time -^- ; 

4 A 


0.3,-^— ' and so on, 


K T 

To the right : a x = d e " 2l ' 4 

_K 3T 

To the left : a 2 = e »*■ ' 4 

_K 5T 

To the right : a 3 = d e »i ' 4 

K 7T 

To the left : a 4 = 6 e ~»i ' 4 

a 2 a 3 a 4 a 5 a n+i 

So that 

Log^=log^...=log^-=g.± = x,say, (18) 

K T 

x = -y , — is called the logarithmic decrement. 

Now we have 

logfe • «t . ?• . . . X -*^) = log (^) n 
\ a 2 a 3 a 4 a »»+iy V a 2y 

Since — 2 = - 3 . . .-^- =-°^ 

a s a 4 a n+I a. 

*. n • log ^i = log ( -^— ) 

Again, since 


OCi Oa . a l _ a l + « 2 

a 2 a 3 ' a 2 a 2 + a 3 

2L* _ <^3_ <*2 _ «2 + a 3 

a 3 a 4 a 3 a 3 + a 4 


. a x 1 , (x.i + a 2 
log — = r lo g * « „ 

«2 + a 3 x a » + a .n+l_ 


w a 2 + a 3 a 3 + a 4 ' ' a n+1 + a 

x== Li og *i + " 2 (20) 

n a n+1 + a B+2 
(i.) The value of x, the logarithmic decrement, may be obtained, 
as in (18), by observing the value of the first swing, say to the 
right, and the successive swing to the left, whence 

x = log, -i, 
oc 2 

or if logarithms to base 10 are used 

x = 2-303 log 10 j 1, 

a 2 

X here depends on the observation of swings. 
It will be seen that a x and a 2 cannot be very accurately observed, 
hence the method of equation (19) may be used. 

(ii.) By observing the 1st and the (n + i)th J swings, the error 

is reduced to - of that in the first method corresponding to 

equation (19). 

(iii.) The third method of finding x, set out in equation (20), 
depends on the observation of swing from left to right (| 
periods). Here no knowledge of the zero reading is required. 
This last method is obviously the one to be recommended, for 
most purposes, of the three discussed above. 

It will be seen that, if we replace <x. 1 + <x 2 by p 1? <x 2 + a 3 = p 8 . . . 
a n + a n+1 = p„ that equation (20) becomes 

>=>£; (2I) 

Since T, the time for a complete swing, is constant when the 
damping is small, we may find X from equation (21) by observing 
Pi directly, and then allowing the system to oscillate a measured 
time until the amplitude is appreciably reduced, say, from half 
to quarter of the original, then measure p B+1 . The number of 
swings between the observations may be found from the time, 
for the number of complete swings may be obtained by dividing 
the time by T; hence (n + 1) is known, p refers to half* a 


complete swing, so that twice the above quotient will give the 
number of swings between observations. 

An application of the above may be seen in the following 
method of finding the logarithmic decrement, which is especially 
advantageous if there is uncertainty in measuring the ends of 
swings, and hence p v 

Suppose we read an even number of successive left and right 
deflections, cc v <x 2 , a 3 , a 4 . . . a m . 

Then Pi = «i + a 2 > 

P3 = «3 + a 4 , 
Ps = <*s + a 6 , etc., 

omitting p 2 , p 4 , etc., for convenience in tabulating. 

After a timed interval, when the deflection is reduced to about 
half the original deflection, obtain the same number of left 
and right readings. From a knowledge of T, suppose the first 
of the second set of readings is a n , and a«, a B+1 , a B+2 . . . 
are observed, we may obtain 

Pn = *»+*n+l» 
Pn+2 = a n+2 ~t~ a n+3> e ^C. 

We have since h = li = hzj = ^ 

P 2 P 3 P» 

Pi _ Pi P 

Pn \U 

and in same way for the other pairs giving 

Pi _ P 3 P 5 

Pn P«+2 Pn 



01 h = Pi + V* + Pi • ■ • = ,*(«-!) 

P« P»+ P« +2 . . . Pn+4 

Hence (»_i)x = log, Pi + P 3 + P 5 • • • 

2-3026 x Pi + P 3 + P 5 ■ „_. , } 

(n-i) gl ° P„ + P B+2 +p a+4 . . . ~l 29) 
The practical details of this are given below. 

Experimental Arrangements 

A suitable form of apparatus with which to make a deter- 
mination of t) for a liquid is seen in fig. 94 

The flat disc is suspended horizontally by a phosphor-bronze 
suspension which is attached to a rod rigidly fastened to the 
centre of the disc. This rod carries a cross-bar whose ends 



have a serew thread, along which two masses may be screwed 
to balance the disc horizontally. 

A small concave mirror is fastened to the rigid support. 

The time of oscillation of this system is first obtained in air. 
This observation is carried out by aid of the usual lamp and 
scale arrangement. A beam of light from a lamp is directed 
on to the mirror, and is brought to a focus by the latter on a 
scale about a metre away. As the spot of light passes its rest 
position on the scale, during its oscillation, a stop-watch is 
started and stopped after 50 complete swings, hence T, the time 
for one swing (i.e. the interval between the spot of light passing 
the zero at consecutive times in the same direction). 

Determination of I. The value of I may be obtained from 
the time of swing. If t is the restoring couple per unit angular 



Fig. 94 

displacement due to the suspension, we have, in air, where the 
logarithmic decrement is small 




The value of T has been obtained as above. 

A small ring of thick copper wire is now placed symmetrically 
on the disc and T 1 , the time of a complete swing, is obtained 
from a determination of the time of 50 swings, for the loaded disc. 

We have 




+ r 


where V is the moment of inertia of the copper wire ring about 
the axis of suspension. 

If a is the radius of the ring of wire, and if the centre were 
in the axis of suspension, I x f = ma 2 where m is the mass of the 
wire ring. 


Squaring (24) and (23) and dividing, we have 

-Z 2 = t + v = ,r_ , ^i 2 

Hence I = T2 ' ma \ 

TV2 <-p2 

r, d, and p may be readily found by the usual methods. 

The logarithmic decrement is now found for the disc oscillating 
in air. Use is made of the method embodied in equation (22) 

The disc is shielded from draught in an empty glass dish, as 
in the figure. A straight vertical wire is placed in front of the 
lamp and by the mirror, and a lens an image of the wire is 
focussed on the scale. 

The disc is given a displacement, and the oscillations are 
observed by means of the light spot. As the spot passes the 
position of rest a stop-clock is started. The first turning point 
of the spot of light is read off, on the right, say, then one turning 
point on the left, and so on, tabulating as under 8 or 10 readings 
on each side. The disc is then allowed to perform the oscillation 
until the amplitude has decreased to half or quarter (depending 
on the time) of the original amplitude. At this stage the reading 
at the turning of the light spot at the right is observed, and the 
stop-clock stopped. The next reading to the left is taken, and 
so on for 8 or 10 on each side. 

The time being t seconds between the starting of the clock 
and the first observation of the second set of swings, T being 

the periodic time of the system in air, there have been — 

complete swings or ^ half periods, hence n as shown under. The 

above process is repeated in the liquid whose coefficient of 
viscosity is to be determined, say paraffin oil. 

Thus, in the table opposite, which gives the observations for 
the disc in air and paraffin oil, we have, if T = 3*07 sec, n, the 
number of the £ swing at the commencement of the second 

set of observations for air = 2 ( — — j = 400, 

\ 3'07/ * 
i.e. for air 

_ 2-3026 An-A 

x °- ( 4 oo-i) logio Wiy 

ss '0027525. 
































Total 411-2 
Time Interval = 615 sees. 



























Total 137-1 


















Total 361-5 

Time Interval = 123 sees. 










Total 91-9 

In the same way the value of the logarithmic decrement in 
paraffin may be calculated, 

2 - 3026 log 10 3 6l -5 

X = 

(n -1) 


n being obtained as above from the time interval measured in 
that case from the commencement of the first set of observations 
to the commencement of the second set. 


Thus, by the methods given, we have a value for I, p, T, r, d, X 
and x ; hence substituting in equation (17) tj may be calculated. 

Repeat the experiment, using water and paraffin oil, and 
calculate the value of y\ for each. 


The variations of the coefficient of viscosity of water with 
temperature may be investigated by means of the apparatus 
shown in fig. 95. 







— - 

■ C 




Fig. 95 

AB is a copper cylindrical vessel into which the capillary tube, 
CD, is fitted by means of a rubber cork. 

Water is heated in a boiler to a temperature 5 or 6 degrees 
above the one at which a determination is to be made, and is 
transferred to AB, the end, D, being closed by the rubber stopper 
shown. The water is stirred and the temperature noted by the 
thermometer, T. It is not very difficult to arrange that the 
water cools slowly and regularly about the desired temperature. 
The water is then allowed to flow into a graduated vessel and 
the time taken for a measured amount to pass. This process 
may be repeated, using water at various temperatures. 

Care must be taken that the temperature, throughout the 
flow, is practically constant, otherwise unreliable results will be 

The radius and the length of the tube may be measured as 


Then, as shown under, we may obtain the value of tj for each 
temperature from the following formula : — 

using the same notation as before. 

In the case of a vertical tube, such as the one described, 
account must be taken of the gravity attraction on the .liquid 
in the tube ; the direct application of the results of equation (4) 
dealing with a horizontal tube being thus inadmissible. 

Consider a vertical tube of length, / cms., and in the liquid 
imagine a thin cylindrical tube of the liquid of radius r, and 
thickness dr. 

Let p be the pressure difference between the two ends of the 
tube, then we have in the steady state of flow pzitr • 8r-\- 2-kt • drlpg 

the downward forces, opposed to -=— • 8r due to viscosity 

where F = yjA • -j-, 


i.e. 2-*r(p + gpQdr = — rf( tjA^V 

or integrating r(p + gpV)dr = — 2-qldv, 

since A = 2twI. 

Hence, once more integrating, 

r 2 

R 2 

where C, the constant of integration = — r(lgp + p), for when 

r = R, v — o, 

i.e- V = (R'-r*)^±^ (25) 

Through such a tube the quantity of liquid passing per second 
s= v (2izr • dr) , 
i.e. the total quantity per second, Q, is 

27W • v dr. 

Substituting from (25) above and integrating, 

7U R* p+gtl 

In the case taken we may apply the result of equation (9), 
page 154, p + %pv 2 — gph — constant. 

If AB is the upper surface of the cylinder which contains 
the liquid (fig. 95), the cross-section being large, the velocity of 



AB will be small compared with that at C. Talcing AB as the 
reference plane for measurement of L, h being the distance between 
C and AB, we have, 

i.e. the pressure, p' , at C is 

P' = n +g?h -|pt> 2 , 
where v is the velocity at the centre of C. 

The pressure at D is n, therefore p = p 1 — n ; 
hence p = gph ~ \pv 2 , 

and thus we have the value below 

O - - JL 4 Z ph -jpfl a +g?l 
* ~ S' 7] I 

Q.-^fl^leL- (26) 

The correction %pv 2 may be obtained from (25) above, i.e. 

To a first approximation, putting p = gph, 
R 2 — r 2 

or at the centred where r — o, 

R 2 gpH 

To the same approximation 

n _ « R* g P H 


Hence we amend equation (26) 

,R* ^ 2Q 2 \ 

This equation is used to calculate -t\ in the experiment already 

Viscosity of Air 

Consider two parallel circular plates, one suspended by a fibre, 
and the other rotating at a constant speed. If the space between 
the plates be filled with any gas a velocity gradient will be set 
up in the layers of the gas parallel to the rotating plate. The 
layer in contact with the rotating disc will move with the latter. 


Due to the viscosity of the gas, the adjacent layer will also 
acquire a velocity comparable with the former. Thus throughout 
the space, the air strata will acquire a motion, just as in the 
case of a liquid flowing through a tube. The layer of the gas in 
contact with the suspended plate will therefore experience a 
force tending to rotate it in the same direction as the constantly 
rotating parallel plate. Due to the force, a couple will act on 
the plate, which will therefore turn through a definite angle of 
such a magnitude that the restoring couple due to the torsion 
in the suspension just balances the displacing couple of the 
viscous drag. 

A & o! Qd b 

R B 

* 1 

Fig. 96 

It can be shown that each stratum of air moves as though it 
were a solid, i.e. it moves as a whole. 

To obtain an expression for the deflection of the suspended 
plate in terms of ij, the coefficient of viscosity of the gas, etc., let 
us assume that the edge effect is negligible, and the gas between 
the plates behaves as though the plates were of infinite dimensions, 
an assumption which is justified by using a guard ring round 
the suspended plate. 

Let d be the distance between the discs, 

w the angular velocity of rotation of the moving plate, 
R the radius of the suspended plate. 

Consider a stratum shown in fig. 96 by the horizontal broken 
line, between EF, rotating, and CD, which is suspended. (The 
parts, AB in the diagram represent a guard ring to eliminate 
the end effect, CD being of radius R.) 

In the stratum considered take a point, P, on a circle of radius 
r cms. about the axis of rotation. 

Points Q and R are the projections of P on CD and EF. The 
velocity of R is r - w ; the velocity of Q is zero. The velocity 

slope is therefore -3-. If uniform, the value of the velocity of 

P is ~ • wr, y being the distance of the stratum considered, below 

the plate CD. 
We have seen that by the definition of viscosity 

F ~ At) Tr ~ " T A ' 


In the stratum considered, let a second circle of radius, r+dr, 
be drawn ; between the two circles there is an annular ring of 
width, dr. The area of the ring is 2nr • dr. The viscous force 
acting on such a narrow ring is, from the equation above, equal 
to F, where 

f = 2^.<*.^t. 

The turning moment about the axis is 


Such a moment acts on the suspended disc on the projection of 
this area. 

Fig. 97 

The total couple is the sum of such couples taken over the 
entire area. Let this couple be C, then 

C = A*r» &dr*= ^ *-* = ^R^. 
Jo & » 4 2d 

Let the suspended plate be turned through an angle, 0, due 
to this couple. The equilibrating couple due to the torsion of 
the suspension is t0, where t is the restoring couple per unit 
angular displacement. This gives 


t9 = ^3- < 2 7) 

A suitable form of apparatus is essentially of the form shown 
diagrammatically in fig. 96. A brass plate is rigidly connected 
to a shaft which may be rotated by means of a belt drive on a 
pulley, which is on the shaft. A small counting indicator 
serves to record the revolutions of the disc, which is steadily 
turned by hand or by a small motor. 

At a distance, which may be adjusted, above the plate is a 
mica disc, suspended by a phosphor-bronze suspension, and 
arranged inside a guard ring of brass. The suspension carries 
a small mirror which serves to measure the deflection by the 


usual method of lamp and scale ; the whole is enclosed in a 
brass case which serves as a shield. A preliminary experiment 
will give an indication of the most suitable speed with which 
to rotate the plate, for any given distance, d, between the plates. 
The zero of the spot of light having been determined, the 
lower plate is rotated until a full scale deflection is obtained. 
The speed of rotation is maintained constant. When this is 
steady, the counting gear, having been read, is thrown into 
action, and a stop-watch started. Maintaining the spot constant 
by steady rotation for as long as possible (at least several minutes) 
the value of the deflection is noted. The counter is then thrown 
out of action and the stop-watch stopped. In that way the 
number of revolutions n, in a known time t, is obtained, hence 

As R occurs in the fourth power, several values of the diameter 
are obtained in all directions, as accurately as possible, and the 
mean value calculated. 

d is measured by means of a cathetometer. From the value 
of the steady deflection and a knowledge of the distance between 
the mirror and the scale, may be calculated in radians. is 
half the value of the angle subtended at the mirror by the length 
of scale, moved over by the spot of light, i.e. tan (20) may be 
obtained, knowing the linear deflection and the distance between 
the mirror and the scale. 

Thus, from this experiment all the terms in equation (27) are 
known except t and 13. 

To obtain t the suspended plate is twisted from its equilibrium 
position, and the simple harmonic oscillations set up are timed. 
If T is the mean value of the periodic time for, say, 50 complete 
vibrations, we have (p. 105) 

= 27t \7 


where I is the moment of inertia of the mica plate about the 
axis of suspension. 

If now the mica plate is loaded by placing on it, symmetrically 
with regard to the suspension, a circle of brass wire of radius 
a cms. and mass m grammes, the moment of inertia has 
been increased to T, where I' = I + ma 2 . 

The time of oscillation, T', of the loaded plate is next deter- 
mined by timing 50 oscillations, then 

- 2 *VT 


Squaring (28) and (29) and subtracting, we have 
T' 2 - T* = 4 TC2 ( r - 1 ) , 
Anhna 2 

T = — . 

^'2 ^2 

Thus, by equation (27) 

2d An 2 ma z t 


7rR 4 (T /2 — T 2 ) 


_ qmaH d 

73 -~" -R4(T' 2 - T 2 ) n ' 
Having obtained the deflection in degrees, <p° say, we have 

= i8o"' 
so that 

tc ma 2 t<p d 


45R 4 (T' -T> 

The Determination of the Viscosity of a Gas (by the Flow though a 
Capillary Tube) 

A simple method for determining the viscosity of a gas which 
may be described as ' The Constant Volume Method,' has been 
described by Prof. A. Anderson (" Phil. Mag." Dec, 1921, 
pp. 1022—3). The apparatus is illustrated in fig. 98. It consists 
of a bulb, V, from which a tube, DO, projects downwards and 
is connected by rubber tubing to a glass tube, AB. These tubes 
contain mercury up to certain levels marked at A and O. 

Just below the bulb a capillary tube leads from the tube, DO, 
as shown at D. This capillary tube is provided with a piece 
of rubber tubing which may be closed by the pinch-cock, P. 

The whole apparatus is mounted on a stand and is of about 
the same size as the constant volume air thermometer. 

The arm, AB, is mounted on a carriage which is readily adjusted 
by means of a rack and pinion which is regulated by the turn- 
screw, C. 

The volume of the gas within V, down to some mark, O, 
and extending to the end of the capillary tube is determined. 
The pinch-cock is then opened to the air and the mercury brought 
well beldw O by properly adjusting AB. The pinch-cock is 
then closed, and the air compressed by raising AB until the 
mercury stands at O. 

A few minutes' interval is allowed to elapse so that the tem- 
perature of the gas, which may have been disturbed in this 



compression, once more attains the temperature of the room. 
The difference in level of the mercury in the two tubes is observed. 
The pinch-cock is opened for a definite accurately measured 
time and then closed. During this interval the level of AB is 
continually adjusted, so that the mercury always stands at O, 
and the gas consequently maintains a constant volume. At 
the end of "the time the difference in level of the mercury is 
again noted. From the observations made, viz, from a knowledge 
of the volume, V, of the gas, the two pressure differences, p x 
and p 2 respectively, the time interval, t seconds, and the baro- 
metric pressure, P, the viscosity is determined from the formula : 

71 =Vt 

ttR 4 

-7- !Og ( 

(*» + P) ifil ~ P) 

8/V ^ ~ 5e (p t + P) {P2 - P) ' 
where R = radius of capillary tube and I = its length. 



Fig. 98 

The above formula may be obtained from an extension of 
the result obtained for the flow of a liquid through a capillary 

Suppose, in the first case, that the gas enters the capillary 
at a fixed pressure P x , and leaves at a pressure P 2 . Let V x and 
V 2 be the volume entering and leaving per second; then 
p x V x = P 2 V 2 . The volume of gas passing any point in the 
tube per second depends on the pressure at the point. The 
velocity is consequently variable along the axis of the tube, 
and therefore the method of the liquid flow cannot be applied 
to the full length of the tube. 

Consider a small element, dx of the tube. Let p be the mean 



pressure in the element and dp the difference in pressure at 
the ends. 
Equation (4), page 152, for the liquid flow, is 

yi ~ l 8 „' 

If q x is the volume of gas passing through the element, the 
above formula may be applied (dx is very small), and since in 

this case — = — ^- (since the pressure decreases with increase 

in x), 

dp it R* 

Now, PjVj = P 2 V 2 = qjp, since the mass of gas passing any 
point is constant, i.e. q x = * * ; therefore 

PiVi^ _dp nJR* 

p ~~ dx'S 7) ' 

F 1 V 1 dx = -~^ pdp. 
Integrating over the limit of the tube 

P 1 V 1 /'W= -Z R * T'p-dp, 

Jo ° 1 J Pi 

i.e. W = ¥ -^r^-T6-T (30) 

If V is the volume available to enter the tube at a pressure P x , 
we may write 


We may apply Boyle's Law to the instantaneous values of 
P and V at entry, 
i.e. PV = constant, 

„dV , „dV . ^dV „dP 

*di +Y dt =°> he - v Tt=~ Y W 

whence from (30) 

(P x » - P 2 2 ) * R* _ _ dP 

/ 16 71 ~ dt ^ 


where -=- is the rate of change of pressure at the end of entry, 

and V the fixed volume. s 

P a in the experiment is the constant atmospheric pressure 


at the end of the capillary, and Pi has values, say, p x and p 2 
at the beginning and ending of the period of observation of 
t seconds duration. Re-writing (31), we have 

ifiWVo* " + 2P J >x V^+P, p,-p J ' 

* R4 * - * rion Pl + p - 2 T a 

i6^V 2P a L g Pi - P 2_L 

~ 2P 2 t 10g £ 2 -P 2 10 \^-Pj» 

which becomes, on writing P 2 = P = atmospheric pressure, 

_/tcR*\ PJ 

71 " \«V J. (p* + P ),( fr - P) 

lg, "¥rrp)(^ 2 -p) 

The term ^%. may be calculated once for all for the apparatus 
and is the ' constant of the apparatus.' 

The Constant Pressure Method 

The determination of yj may be made with constant pressure 
difference between the ends by maintaining the levels of the 
mercury columns in the two tubes at a fixed difference. The 
point, O, would be chosen at the commencement near the bottom 
of the tube, and the time taken for a volume corresponding to 
a length of tube between the original and final positions of O 
measured. If p is the total constant pressure inside V, we have 

v » = tf l6 V (30) ' 

in which all terms except tj are known. 

The difficulty in this modification lies in keeping the mercury 
levels a fixed distance apart as the gas is driven out. 


The Comparison of a Thermometer by means of a Standard 

It is recommended that each student before beginning his 
experiments on heat should choose a thermometer, test its 
accuracy, and use it when required throughout his experiments. 
The method of comparison is to immerse it together with a 
standard thermometer in a bath and observe temperatures over 
a suitable range simultaneously by both instruments. 

It is convenient for the following experiments to have two 
thermometers, one reading from o° C. to 35 ° C. and the other 
from o° C. to ioo° C. Both should be calibrated in this way. 

A large water or oil bath should be carefully heated over a 
Bunsen flame and constantly stirred. The thermometers should 
be placed in the liquid so that the mercury thread shows just 
above its surface, and with their bulbs close together. 

A record of temperatures at intervals of 5 should be taken 
over the range, and a curve drawn with temperature corrections 
as ordinates and with readings from the thermometer to be 
calibrated as abscissae. 

By means of this curve the readings of the thermometer in 
later experiments can be reduced to that of the standard. 

A better method of heating the liquid is to place it in a bath 
standing in a box lined with cotton wool and to supply the heat 
by passing an electric current thrbugh a resistance coil immersed 
in the liquid. If the current is drawn from storage cells, and a 
variable resistance included in the circuit, it is possible to adjust 
the current so that the bath is maintained for a long time at a 
constant temperature. The most convenient form of stirrer is 
a small propeller driven by a small electric motor. 

The Calibration of a Mercury Thermometer 

It is impossible in practice to obtain a perfectly uniform bore 
in the stem of a mercury thermometer, so that it is not sufficient 
in an accurate instrument to divide the interval between the 
fixed points into a number of parts of equal lengths. The 
makers of thermometers usually attempt to make some 
correction for this lack of uniformity by adjusting the distance 



between consecutive divisions to suit the bore of the stem at 
the various points. But, in spite of this, unless the thermometer 
is exceptionally carefully constructed, errors remain and a cali- 
bration has to be made if accurate observations are required. 

In a given thermometer, as a rule, the divisions will be unequally 
spaced at different parts of the tube, and the bore will vary from 
point to point. 

The first step is to divide the tube into segments consisting 
of five or ten degrees each, and over each to find the average 
distance between each division. We assume for the sake of 
definiteness that we are considering intervals of ten divisions. 
Measure each of these, beginning at o° by means of a micrometer 
microscope, and deduce the average length per degree for each 
of the ten intervals up to ioo°. We are assuming that the 
thermometer is divided into degrees Centigrade from o° to ioo°. 

When this has been done a thread of mercury is broken off 
from the main column of length equal to that of about io° on 
the thermometer scale. 

This thread may be obtained by connecting a small jet, made 
by drawing a glass tube to a narrow bore, to a gas-pipe, and 
lighting the gas at the narrow end, adjusting the supply to 
produce a flame about half a centimetre long. If this flame be 
applied cautiously at the point of the thread where it is desired 
to sever it, the thread will divide. During the application of 
heat the thermometer must be rotated to avoid fracture due to 
unequal heating. 

The thread is moved by gently jerking the thermometer until 
one end is at o° and the other near io°, and its length measured. 

The thread is then moved so that one end lies at io°, and the 
other near 20 , and so on up to ioo°. 

If it is difficult to get the detached thread down to zero, 
owing to the projection of mercury from the bulb past the zero 
mark, the bulb should be cooled by wrapping it in wool and 
moistening with ether. This will clear the tube and allow the 
thread to be moved down without its re-joining the main mass 
of mercury. 

Denote the ten lengths of the thread by l lt Z 2 , l z , etc. 

Let these be reduced to their equivalents in degrees. This 
is readily done since the average width of one degree is known 
in the different parts of the scale. Denote these equivalent 
lengths by t x , t 2 , t z , etc., and the mean of these by t. 

If the tube had a uniform bore of the same length as that of 
the actual instrument between o° and ioo°, the readings would 
be t, 2t, 3*, etc., instead of t lt t x + 1 2 , t x + 1 2 + t z , etc. 

Let the values added respectively to t x , t 2 , t 3 , etc., to make 
them equal to / be d v 6 S , <S 3 , etc. 



t = t x + 8 X 
t = t z + <5 a , etc. 

The corrections to be applied near io°, 20 , 30 , etc., are therefore : 


t-t x , 

2t — t x — t 2 , 3* — h — h — *a> etc., 

<5i + d 2> 

^1+^2+ 5 3- 

The correct temperature corresponding to t x is (t x + d x ), and 
corresponding to the point [t x + 1 2 ) it is {t x + t 2 + 8 X + <5 2 ), 
and so on. 

We have assumed up to the present that the fixed points 
at o° and ioo° are correct. 

A table is drawn up as shown below. The correction to be 
applied near io 9 is — -023, for the thermometer column is too 
long at this point, near 20° the correction to be applied is 
(_ .023 + -032) or + '009. The correction would have been 
+ -032 had the point near io° been correct, but since that was 
not the case both errors come in. We correct similarly for other 
points by adding errors algebraically, and recording in the last 
column the amount to be added to the thermometer reading to 
obtain the correct temperature. 

Note that in the example given the thread was not very near 
to the mean length of io°. It is convenient to arrange this as 
closely as possible to io°. Strictly, the error —023 ought to 
be applied to the recorded temperature of 10-231, but the error 
will probably not vary very rapidly in the neighbourhood of any 
given point. Hence we take the corrections in the last column 
as applied at io°, 20 , etc. 

Mean length of thread, as deduced from column 4, 10-208. 











O - IO° 
IO°- 20° 
20°- 30 
3O - 40 
4O - 50° 

50 - 6o° 
6o°- 70 
70 - 8o° 
8o°- 90 
90°-ioo Q 





8 X — 023 
<5 a +'°32 
<5 3 —035 
<5 4 +-oi5 
<5 6 +-041 
6* —"5 
0- -ooo 
<5 8 +-038 
6 9 +-016 


+ .009 

— •026 

— •Oil 
+ •030 





+ •004 


Draw a curve with thermometer readings as abscissae and the 
corrections to be applied to obtain the corrected readings as 
ordinates. The curve should pass through the #-axis at o° and 
ioo°, since these points have been assumed to be correct. 

If, however, the fixed points are incorrectly placed, the errors 
must be found in the usual manner with ice and steam. 

Suppose that the zero correction is d , while that at ioo° is 


The upper fixed point must be corrected for pressure, latitude, 
and height above sea-level. 

The barometer must also be corrected, owing to the fact that 
it is probably not read at the temperature at which the instrument 
was standardized. 

This last correction may be made by the following formula : 

If h denotes the height at o°, and h t that read at t°, 

h = h t {i — -0001622). 
For latitude x and at a height d feet above sea-level, the 
length of the column which produces the standard pressure at o° 
and at sea-level in the standard latitude of 45 ° is : 

L = (760 + 1-94562 cos 2X -f- -000045466^) mm. 

Under this pressure, L, the boiling point is ioo° at the height 

and latitude of the place of observation. 

h h 

Thus, h is equivalent to -j- atmospheres or ^- x 760 mm. 

under normal conditions. From this and the following table 
the correction to the boiling point may be made. 




720 mm. 

98-493° c. 













755 , 


760 1 










The zero on the scale actually records the reading 
the ioo° records (100 — 5 100 ). 
Thus a correction is required for this. 

d n , and 


Plot on the curve the two points (0,0), (100, <5 100 — <5 ), 

and join them by a straight line. The ordinate of this line at 

x° is (3 100 — <5 ) • — , since each degree, even if correct as 

regards bore, would register only = — 10 ° ° , we must add 

J 100 

to it an amount, — — , on account of the errors at the 


fixed points 

To x° we must add 


Thus, if we draw on the same graph on which the first set of 
results was plotted this second curve the difference between the 
two ordinates taken algebraically will give the true reading 
corrected for errors due to bore and fixed points. 

Re-draw a new curve, showing as abscissae temperatures as 
recorded by the thermometer and as ordinates the difference 
of the orcjinates of the two curves, and the resulting curve will 
give the amount to be added to any recorded temperature to 
give the true temperature. 

Throughout the taking of measurements the temperature of 
the detached thread should remain constant ; and in order to be 
sure that this condition holds, place another thermometer close 
by and observe whether it varies or not. Do not handle the 
thermometer under examination more than is necessary, and 
only do so by holding it at the tip away from the bulb. 

Newton's Law of Cooling 

The object of this experiment is to verify Newton's Law of 
Cooling, which states that the rate at which a body cools is 
proportional to the difference of temperature between itself and 
the enclosure in which it is placed. The constant of proportion- 
ality depends on the surface exposed and the thermal capacity 
of the exposed body, and the law is true for small differences of 
temperature only. 

The apparatus required is a small metal thimble, into which 
water at about 8o° C. can be placed, provided with a cork through 
which a thermometer passes for noting the temperature of the 

The enclosure consists of two calorimeters, one inside the 
other, and containing water in the space between, to provide 
an enclosure at nearly constant temperature. A thermometer 


placed in this water, which should be stirred occasionally, gives 
the temperature during the experiment. 
Observe the temperature recorded by T x and T a (fig. 99), at 

Fig. 99 

intervals of half a minute during the initial stage of the fall, 
and, as the rate decreases, the interval between observations 
may be increased. The record of T 2 should not vary very much. 
Tabulate the results thus : 




72-5° C 

12-2° C. 


7i-3° C 

12-2° C. 


70-1° C. 

12-2° C. 


69-0° C. 

12-2° C. 


67-8° C. 

12-2° C. 


667 C. 

12-2° C. 


65-6° C. 

12-2° C. 


6 4 -5° c. 

I2'2° C. 


63-5° c. 

12-3° C. 


62-5° C 

12-3° C. 




Draw a curve showing the relation between the temperature 
Tj and the corresponding time. 

Make the temperatures the ordinates and times the abscissae. 

The rate of fall of temperature may be obtained from this 
curve by measuring the tangent of the angle of inclination of 
the tangent to the curve to the axis of t. As explained in the 

introductory chapter this measures the value of -^ at the 

various points of the curve* 



According to Newton's law these values should be proportional 
to the differences between T x and T 2 . If T 2 does not vary very 
much its mean value may be regarded as the mean temperature 
of the enclosure. x 

If T 2 varies too much to permit this approximation, draw a 
curve showing the relation between T 2 and the time on the 
diagram which shows the relation between T x and the time. 

Then at each time we can determine the value of (T t — T^ 

and the corresponding value of -— from the same graph. 

Make another table containing two columns, one for values 
of (T x — T 2 ) and the other for the corresponding values of — j^- 

Draw a curve with the values of (T 2 — T 2 ) as ordinates, and 
those of -~ as abscissae, when, if these two quantities are 
proportional, the result should be a straight line. 

The Use of the Weight Thermometer 

The weight thermometer consists of a glass bulb, B, drawn 
out at the upper end into a capillary stem, A. 






Fig. ioo 

A convenient size is obtained by making B about 6ne and a 
half to two inches long, and between a quarter and half an inch 
wide. It forms a good exercise to make the apparatus from a 
piece of glass tubing. If this is done, care must be taken to 
get rid of any blob of glass likely to accumulate at the bottom 
of B, otherwise in the subsequent heating B will be very likely 
to crack. 

The apparatus measures the expansion of any liquid placed 
in B relative to glass. In order to deduce the real expansion it 
is necessary to know that of B. 

The coefficient of expansion of mercury has been found with 
great accuracy, and may be taken as -0001818. 


We may therefore use mercury to find the expansion of B, 
and then use B to determine the expansion of other liquids. 
We shall consider the determination of the expansion of water. 

(1) Determination of the Expansion of the Weight Thermometer 

Carefully weigh the apparatus. 

Surround B by wire gauze and warm carefully in order to 
drive out air. Place the end A under clean mercury contained 
in a dish and allow B to cool so that a little mercury enters. 
Then, when sufficient has entered, boil the mercury so that the 
space above it may become full of mercury vapour. Once more 
place A below the mercury, when B will fill as the vapour 
condenses on cooling. 

It is best to carry out the process gradually, heating and 
cooling B several times, and allowing a little mercury to enter 
at a time. 

Warm up the mercury first to prevent the bulb from cracking 

when it enters. 

Allow B to cool gradually, and finally surround it with ice, 
keeping A immersed all the time. 

Bring up a small weighed, empty dish, and remove the mercury 
at A, replacing it by the empty dish. Remove the ice and allow 
B to acquire the temperature of the room. Mercury will, of 
course, flow over into the dish. As soon as the flow ceases, 
weigh the bulb, B, and the dish. Let the total weight of mercury 
within B at o° be denoted by W . 

Immerse B and as much as possible of the stem in boiling 
water, and catch the mercury that flows out in the little dish. 

After the bulb has remained in the water for a quarter of an 
hour, to allow it to assume the temperature of the boiling water, 
remove it, dry and carefully weigh it. Suppose that w is the 
amount that has flowed out. As a check re-weigh the dish and 
again determine w. 

Let p denote the density of mercury at o° C. and p t that at 
temperature T. 

If V denote the volume of the apparatus at o° and V x that 
at T, while p denotes the coefficient of expansion of glass, 

£ = * + P T. 

But V =-^, 


- 1 + Pi - Wfl pi W() l^^' 


• s =, W ° ~ w * +ocT x 
" W ' T T 

W — ze> te> 

— a — 

W * W T 
« =H7^ P + * 


W — w W Q —w T 

W W I 

p ■ 

W -a>; p T W -ze> r 

The ratio ^ _ ^ is small, and since p is also small we may 
often neglect the first term on the right and use simply 

a - p = w ,1 . 

P W -ze> T 

(2) The Coefficient of Expansion of Water 

The value of this coefficient varies considerably throughout 
the range o° to ioo°. In this experiment the average value 
between two temperatures, say, 20 C. and 6o° C. will be deter- 
mined. If both temperatures are above that of the room the 
experimental difficulties are not so great. We assume this to 
be the case though the method is quite general, and exactly the 
same precautions have to be taken as in (1) if water flows out 
on moving B from the lower temperature enclosure to the balance. 
Carefully fill the weight thermometer with water that has been 
recently boiled to get rid of contained air, and place it completely 
immersed in water at temperature t x . Remove it, dry, and weigh. 

Place the thermometer in water at t 2 and repeat. 

Use the formula given above to deduce a for water, making 
use of the value of p previously determined. 

Determination of the Density of Water at Various Temperatures 
by means of a Glass Sinker 

In this experiment a solid is weighed while totally immersed 
in water at different temperatures, so that by the principle of 
Archimedes the weights of the fluid displaced by the solid 
corresponding to the various temperatures are known. 

Let V denote the volume of the solid at o° C. and a the co- 
efficient of expansion of the solid, so that at temperature t° 
the volume is V (i + at). If Pt is the density of the water at 
temperature, t, the loss of weight due to immersion is 
V (i + a*) p t . This value is observed by the balance ; let it be 

W«. Then p ,=- — ^i 


In order to carry out the experiment, a long Wire is attached 
to the scale-pan of a balance and passed through a hole in the 
base of the balance to support the solid, which hangs in the 
water. Since during some part of the experiment the water 
will be at a temperature considerably above that of the immediate 
surroundings of the balance, it is necessary to use a wire about 
40 cms. long, so that convection currents may not disturb the 
equilibrium of the balance. 

The wire should have a diameter not greater than T V mm., so 
that surface tension may not cause any appreciable effect where 
it enters the water. In practice thin copper wire is often employed, 
though it is preferable to use a short length of platinum wire, 
specially treated to diminish surface tension effect for immersion 
in the water. ' # 

A stirrer is necessary to keep the temperature of the liquid 
uniform, and as soon as the whirls, due to stirring, have died 
away, the balance is made and the temperature taken by a 
thermometer immersed in the water with its bulb as near as 
possible to the solid. 

It is preferable to heat up the water to the highest desirable 
temperature and allow it to cool down. In this case the weights 
in the scale-pan will require to be continually diminished. The 
weights should be adjusted before an observation so that the 
solid appears a little too heavy. After a short interval the scale 
pointer will cross the zero position, and at this instant the tem- 
perature of the water should be observed. 

In this way a series can be obtained very conveniently for 
temperatures above atmospheric. 

The solid is first weighed in air so that the values, W*, corres- 
ponding to different temperatures may be obtained by 

As a sinker it is usual to employ a glass bulb containing lead 
shot, and a may then be taken as -000025. 

When it is not possible to reduce the temperature right down 
to zero it may be taken down to some convenient low temperature, 
t . By finding the weight, W tQ , of the submerged vessel at this 
temperature, the volume, V tQy may be deduced by the help of 
the density table appearing on p. 188, taken from Kohlrausch's 
" Physical Measurements," 

v - W ">. 

since V '»~7r 

The densities at the other temperatures may then be deduced 
by the formula 


p < V t0 {i + «(*-g}' 




























16 6 




















In the laboratory it is convenient to begin at a temperature 
of about 8o°, and make observations about every io° C. 

A curve should be drawn exhibiting the relation between 
temperature and density. 

Alternatively the experiment may be made in order to deter- 
mine the coefficient of expansion of water for varying intervals 
of temperature from the formula : p = p*(i + p*). 

The Constant Pressure Air Thermometer 

In this form of thermometer temperature is denned by means 
of the equation 

V< = V (i + at) ; 

t denotes the temperature, V t and V the volumes occupied of 
a certain mass of gas at two temperatures, the former at t°, and 
the latter at a convenient fixed point : the zero of the scale. 

By choosing the melting point of ice as t = o°, and the boiling 
point of water as t = 100 °, under standard conditions we can 
find the value of a. 

We may therefore say that the equation is assumed, and that 
t is defined by it, 



It is assumed that the pressure remains constant throughout. 
The diagram shows a simple form of constant pressure thermo- 


The mercury reservoir is adjusted so that the mercury stands 
at the same level in the two tubes, CD and EF. When this 
is the case the pressure is equal to that of the atmosphere in 
both tubes. 

Suppose it is desired to determine a certain temperature with 
this instrument, say the melting point of wax. First surround 
the bulb completely with powdered ice and turn the three-way 
tap, T, so that B and CD are open to the room, and adjust G 
so that the mercury stands at the zero division on the graduated 
scale of CD. 

Fig. ioi 

Wait for about ten minutes to allow the bulb to cool exactly 
to o° C, and turn the tap so that B and CD are connected to 
each other but cut off from the atmosphere. In this way the 
zero on the scale is made to correspond to o° C. 

Now immerse the bulb in boiling water, and lower G until 
the mercury stands at the same level in the two tubes, and 
observe the scale reading. 

Read the barometric height, and deduce the boiling point of 
the water. 

Surround the bulb with warm water and adjust its temperature 
to the melting point of the wax. To do this, put a small piece 
of the wax in a small test tube and immerse it in the water. 
Heat the water until the wax melts, and then let it cool a few 
degrees, and then warm up very slowly, keeping the water 


stirred until the wax begins to melt again and then take the 
reading of the thermometer. Of course G must be adjusted so 
that the level of the mercury is the same in both tubes. 

From the observations made we can deduce the melting point 
of the wax. Note the temperature also by means of an ordinary 
mercurial thermometer. 

Theoretical Considerations 

The fundamental equation of gas thermometry, whatever may 
be the form of thermometer, is simply : 

Total mass of gas in the instrument = constant. 

In practice it is not possible to maintain all the gas at the 
same temperature ; some of it is necessarily remote from the 
point of application of the body examined. These remote 
regions are described by the term ' dead space.' 

In our apparatus the dead space extends from fl to the level 
of the mercury in C. 

We shall suppose that the scale readings are in ccs., beginning 
at the zero and extending downwards. 

Let the reading corresponding to the case when boiling water 
surrounds B be denoted by x b , and let x w be the reading when 
the wax is melting. 

Let the volume from the top of the bulb at the point where 
it is immersed to the zero of the scale be denoted by v, with a 
suffix to indicate the temperature at which it is measured. 

The temperature of the dead space, which has a total volume 
(v -f x), will vary from one end to the other ; but we shall make 
our calculations by assuming that this temperature is uniform 
throughout and equal to that measured by placing a thermometer 
in a position approximately midway between the two ends of® 
this space. 

We shall denote this by the letter t, and when the bulb 
temperature is t we shall write t t for the corresponding tempera- 
ture of the dead space. 

Let the volume of the bulb together with that part of the 
tube which is immersed be V at the temperature zero, let p 
denote the coefficient of cubical expansion for glass, and p the 
density of air at zero. 

The mass of gas in terms of the quantities measured when the 
bulb is at o° C. is : 

V Po + «V Pr • 

The temperature of the dead space is, of course, not necessarily 
at o°, it has some value t . 

Or we may write for this mass : 



T -*SHH3) '■' 

When the bulb is at a temperature, t, let x t denote the reading 
on the scale. This denotes the volume between the zero of the 
scale and the mark x t at the temperature at which the apparatus 
was graduated. This is often done at about 15 C, and the 
laboratory temperature is usually in this neighbourhood. We 
shall not introduce any great error into our calculations if we 
regard this as measuring the true volume of this part of the 
apparatus under the conditions Of the experiment. 

The temperature of the dead space is now t t , and the total 
volume is therefore : 

v (i + p«r t ) + *t 

The mass of gas in the dead space is : 

v (z + PT f )p x tPo 


I + OCTf 1+ OCT 

Hence the total mass is measured by : 

V (i + Pflpp , v (i + Pt,) Po x tPo 

i+att "*" 1 + <xt, "t" 1 + <xt, {2) 

If the expressions (1) and (2) be equated, since they denote 
the same quantity, it will be found that : 

*. = v o(I + .Tof^i* + f«{i±^ -f±^n . . o) 

l_I+otf V (i+aT 1 + aTjJ w; 
The second term in the square bracket may be written : 

?o(« - P) ( T f - T o) 
V. (i + oct ) (1 + ax t )' 

where we have neglected the product a p in the numerator. 
In practice the difference between t t and t is not very large, 

and the construction of Jhe apparatus provides that ^is small. 

We may thus neglect this term in comparison with the first 
term in the bracket without introducing any great error. The 
student is recommended to find an approximate value of the 
two terms, so that he may better appreciate the effect of this 
neglect of the second term. 

We then have : 

- *• = v «<« - «« • TT#- (4) 

If we apply this to the case when the bulb is surrounded by 
boiling water, of which the temperature is b, corrected of course 


for any variation of the barometric height from normal, we 
have from (4) 

*» - V (a - p)6 Z + aT& 

Hence from (4) and (5) 

1 + ub 

I + a6 £ 



I + a^o I + orf 

-» + T - lr + » , 

a a 

The value of — may be taken as 273*1. 

j t and t h may be observed on a mercury thermometer, although 
this is introducing into the experiment the mercury scale. 

Equation (6) is a linear equation in t, which is thus determined 
from the reading on the scale of the air thermometer. 

This enables us to deduce the temperature of the melting wax. 

Fig. 102 

The Constant Volume Air Thermometer 

The diagram illustrates a common form of the apparatus. 
The bulb, B, is connected by a capillary tube to rubber tubing, 
DE, and to the glass tube, EF. 


EF slides against a scale, SS, by means of which the height 
between the mercury levels in the tubes, CD and EF, can be read. 

EF can be clamped in any desired position, so that the level 
of the mercury in the tube CD stands always at a definite mark, C. 

In this case the temperature is denned by assuming the relation : 

p t =p (i+<rt), or t=l.t±Zl±*. 

p t denotes the pressure within B at a temperature t, and po 
that at a standard zero position — the temperature of melting 
ice, while the volume of gas remains constant. 

The pressures are measured by adding the atmospheric pressure 
to that due to the mercury column of length, h. 

In practice the part of the apparatus containing the air expands 
with rise of temperature, and there is the dead space, HC, to be 
alloweoVfor as in the last experiment. 

Let V denote the volume of the bulb and immersed portion 
of the apparatus, v that of the dead space at o°. 

Suppose that the dead space remains at temperature t during 
the experiment when B is surrounded with ice, and t& when it 
is surrounded by steam. 

We shall show how a may be determined experimentally. 

If p denote the density of air under normal conditions, and 
p that at pressure, p, and temperature, t, 

_ Po P_ 

9 - I + at ' 76 
The mass of gas contained in the thermometer is : 

-^-•V + 76 ^o (l+aTo) (7) 

expressed in terms of the conditions prevailing when B is at 
the temperature of melting ice. 

Similarly the mass of gas expressed in terms of the conditions 
prevailing when B is at temperature, t, is : 

Vq(i + PQ-/>iPo ■ Vo(t + t*t)P#o ,™ 

76(l+0C/) ^ 76(1+ OCT,) W 

Hence on equating (7) and (8) : 

Again on account of the smallness of the ratio ~ we have as 

an approximation : 

*-*'T+* (I °) 

J 3 


When the bulb is at the temperature, 6 i of boiling water, 

i + $b 

Po =Pb 

I + <x& 


We may assume the value -0000232 for p and thus calculate a 
from (11) by observing the pressures when the bulb is surrounded 
by melting ice and by steam respectively. 

Equation (10) then enables us to deduce the temperature 
corresponding to any pressure, fi t . 

Take a mercury thermometer and immerse it close to the 
bulb, observing its readings and the corresponding pressures. 
From the latter deduce the temperatures from (10) and draw 
up a table recording these in one column opposite to the records 
of the mercury thermometer in a second column. 

Draw a graph with air temperatures as ordinates and mercury 
temperatures as abscissae, exhibiting the deviations between the 
two temperature scales. 

In order to calculate the value of —^ if this is necessary, first 

adjust the mercury to the mark C, immersing B in watei that 
has come to the room temperature. 

Fig. 103. 

Read off the pressure, P, to which the air is now subjected. 

Carefully raise EF so that the mercury approaches the bend 
at H, and so fills nearly all the dead space. Let the pressure 
within the bulb be now P.. Then, since the conditions are 


isothermal and the whole volume, v , has been very nearly filled 
with mercury in the second case, 

P 2 V = PxCVo + v ), 

*V P x 

We may read the temperatures t and t t by means of a mercury 
thermometer placed close to the dead space and obtain a closer 
approximation to the value of, a. 

In the second form of apparatus (fig. 103), into which the first 
may be readily converted, the volume of the dead space is made 
negligible. _ 

In order to measure the difference of level between C and F, 
two tubes, WW, connected by a rubber tube containing water 
are adjusted so that the level on the left is the same as that 
at C, and consequently this is the same on the right at C 1 . 

The distance, C*F is readily observed on SS. 


The Specific Heat of a Solid by the Method of Mixture 

The student will be familiar with the principle of the method 
of mixture. The main object in this description is to give an 
account of the method of making a correction for the error 
arising from radiation. 

If W is the water equivalent of the calorimeter and contents, 
m the mass of the solid, and s its specific heat, and if t x is the 
initial temperature of the calorimeter, t % the final temperature 
and T that of the solid initially, then if there has been no loss 
of heat we have : 

ms (T - jQ = W(* 2 - t x ). 

In practice there is a loss or gain of heat from or to the calo- 
rimeter, which should be added on the right-hand side of this 
equation, since all the heat from the solid has not been retained 
in the calorimeter. 

We may make the correction by the consideration that the 
final temperature, t 2 , would have been t 2 + At, where At is an 
interval of temperature which must be small if the experiment 
is to be successful. 

Hence the corrected equation is : 

ms{T - t 2 ) = W(t 2 + At - t x ). 

At will be small if during half the experiment the calorimeter 
gains heat, and in the other half loses heat. This can be arranged 
by adjusting the initial temperature so that the room temperature 
is approximately a mean between it and the final temperature. 

A preliminary experiment is made to find out roughly the 
temperatures that will be attained during the experiment. 

The same amounts of the materials are used in a second case, 
but the calorimeter is cooled down by adding small pieces of 
ice or warmed up, as may be necessary, so that t x and t 2 may 
lie at nearly equal temperature intervals below and above the 
temperature of the room. 

The temperature of the calorimeter is noted immediately before 
immersing the hot body, and then at quarter or half minute 
intervals until the maximum temperature is attained, the 




observations being continued beyond this point at definite 

These results should be plotted on a graph (fig. 104). 

The curve obtained will be similar to ABC. Had there been 
no losses or gains on account of radiation, the curve would have 
been similar to ADE, the final -temperature remaining constant 
at the level DE. 


Fig. 104 

We can, by applying Newton's Law of Cooling, derive the 
curve ADE from ABC. 

Suppose the axis of t to be divided into small intervals, OM x , 
MjMji, M 2 M 3 , etc., of magnitudes 8t x , 6t z , dt z , etc., and let the 
average temperature during these be : X , 2 , 3 , etc. 
Then the loss by radiation in 8t x is kd x 8t x , where k is a constant. 

.-. p^i = QxMi + ke x dt x . 
Similarly the loss during MiM 2 is kd 2 8t 2 , and consequently 
P 2 M 2 = Q 2 M 2 + kd x dt x + &0 2 # 2 . 

This process may be continued to any extent, and if we make 
the intervals sufficiently shorty M x , Q 2 M 2 , etc., are not sensibly 
different from 6 X , 2 , etc. 

Thus, denoting by © the ordinate of the upper curve at a 
time, t, and by that of the lower at the same time : 
©! = ©! + ke x 8t x , 
© 2 = 2 + ke x 8t x + kd 2 8t 2 ; 
and generally : 

© rt = 0„ + k{6 x dt x + 2 # 2 + . . . + B n 8t n ) 

= n + &(area of lower curve from OA to ordinate B ), 


© = + k I ddt 


We can thus make the correction by carefully drawing the 
lower curve, calculating the area up to the point 0, which it is 
desired to correct, and add its product by k to 0. 

We require in this experiment the ordinate corresponding to 
any point P 1 along DE, beyond the point B. 

We therefore determine graphically OAQ x M and apply the 
correction to Q X M, thus obtaining P*M. 

In order to calculate k, find th6 rate of cooling along BC, 
corresponding to a mean temperature 0. 

k = — • j- (Newton's Law.) 

It will usually be convenient to note the fall for four or five 
minutes, and make the deduction from it. 

Reynault's method of making the correction is actually to 
make the correction of the ordinates and draw the curve ADE. 

For this purpose we require to know the rate of cooling at any 
particular temperature. 

Find the rate at one period, as above, and plot jr against 

on a curve. In doing this it is assumed that the relation is 
linear in accordance with Newton's Law. 

At the temperature of the room ~ vanishes ; for if the calo- 


rimeter and surroundings were at the same temperature then 

there would be no loss due to radiation. 

Fig. 105 

We therefore have two points on the graph, and by joining 

these by a straight line we can determine -j- at any value of 6 


(fig- 105.) 

Divide ABC into sections very nearly straight, as AR, RQ, etc. 

Note the mean temperature over AR and from the graph for-r 

note the rate -5- for this temj^erature. Multiply this by the 



time, ON, and add the result to the ordinate, NR, thus obtaining 
NR 1 . Let this correction be <50 v 

In the same way find the amount <50 2 , lost during the interval, 
NM. Add the sum (<50 x -f <50 2 ) to MQ and so obtain MQ 1 . 

Continue this process until the maximum ordinates along DE 
are attained. 

The curve obtained in the experiment should attain this 
horizontal branch very nearly, and the ordinate is the quantity 

Speciflc Heat of a Liquid by the Method of Cooling 

The rate of loss of heat of a body depends only on the tempera- 
ture of the body and that of its surroundings, on the area, and 
on the nature of the surface exposed. 

If the difference of temperature between the body and its 
surroundings is not large, the rate of emission of heat is propor- 
tional to the temperature difference. This is Newton's Law of 

Suppose a mass of liquid, M lf is enclosed within a calorimeter 
of mass m x , and let S x and s x denote the specific heats respectively. 
The thermal capacity of the system is (M^ + m^^. If the 
temperature fall from t t to t 2 in n x seconds, the average rate of 

loss of heat is (MjSj + m x s x ) . — - 

In the case of the second liquid under the same conditions, 
let n 2 denote the number of seconds required for a fall of tem- 
perature from t x to t 2 , and the loss of heat per second is 


We have by Newton's Law 

(M X S 1} + m x s x ) 


= (M 2 S 2 + m x s x ) 


:. s x = 

n x (M 2 S 2 + m x s x ) 


_ m x s x 

n 2 M x "M7 

The apparatus for carrying out the determination consists of 
a small calorimeter fitted with a rubber stopper through which 
a thermometer may pass (fig. 99.) A calorimeter of aluminium 
of about an inch diameter and three inches high serves the purpose 
very well. The calorimeter should be supported by threads, or 
should stand on a non-conductor within a double-walled enclosure, 
the thermometer passing through a cork in the lid of the enclosure. 

In order to secure a uniform temperature, the space between 
the walls of the enclosure may be partly filled with water. In 
this case care must be taken that the inner box does not float, 
or it may happen that it will touch the calorimeter and there 
will be loss of heat by conduction. 

The enclosure may consist of two calorimeters — an outer large 
one fitted with a lid, and an inner smaller one standing on blocks. 

First fill the aluminium calorimeter about two-thirds full of 
water, and warm it to a temperature about 70 C. by immersing 
it in hot water. Place the apparatus in the position shown in 
the diagram, and take readings of the thermometer at intervals 
of half or whole minutes down to a temperature below 30 C. 

Note from time to time the temperature of the enclosure, 
which should hardly vary during the experiment. 

Fig. 107 

Some time must of necessity elapse between the observations 
on one liquid and those, on another, and although it is not difficult 
to maintain a constant enclosure temperature throughout each 
set of observations, it often happens that the mean temperatures 
recorded by the thermometer, T 2 , are appreciably different in 
the two cases. 



This difficulty may be avoided by using a second aluminium 
container similar to G, and suspending it by the side of G inside 
C. The records of the temperatures of the two liquids are then 
made almost together, and the enclosure temperature is the 
same for each. 

Make up a table containing the liquid temperatures opposite 
the times of observation, and in a third column record the 
enclosure temperatures. 

Draw on the same graph as illustrated in fig. 108 the curves, 
one for each liquid, with the differences of temperature between 
liquid and enclosure as ordinates, and with the times for abscissae. 

Let AB denote the curve for paraffin (say) and A B that 
for water. 

Draw the horizontal lines, T X AA and T a BB to cut the curves 
at A, A and B, B as shown. 

Let t x denote the temperature of the liquids above that of 
the enclosure in the first case, and t 2 the corresponding 
temperature in the second. 

Then in the time that elapses between the instant measured 
by TjA until that measured by T 2 B the paraffin cools down the 
interval (f r — t 2 ), and the water cools down the same amount 
during the interval between T^o and T 2 B . Denote the two 
periods of cooling by n x and n z . In this case the value of S a is 
unity, and for aluminium the value of s x is -219. 

By weighing the liquids and calorimeters we obtain sufficient 
data to give the value of s 2 for paraffin by means of the formula (1). 

Determination of the Specific Heat of a Solid by means ol Joly's 
Steam Calorimeter 

A metal jacket, J, enclosed in a casing of felt surrounds a 
platinum pan, P, suspended by means of a fine wire attached 
to one arm of a balance, whose base is shown at BB, 


The upper end of J is closed by a light metal disc, D, through 
a hole in which the wire passes. This disc is free to move, 
and when oscillations occur in P it finally settles down so that 
the wire passes through the hole without contact with the disc. 
Just above D, a small coil of wire carrying a heating current 
round the suspension prevents condensation on it and also on D. 

Fig. 109 

In the first place, J is allowed to attain the temperature of 
the room, and the inlet and outlet pipes are then closed. 

The pan is balanced in the usual way, and in the meantime 
water is boiled in a container ready to supply J with steam by 
means of I. 

When a good supply is obtained O is opened and steam passed 
through I. When the steady state is attained it will be found 
that additional weights are required to counterpoise P on account 
of the condensation of steam on it. Suppose w grammes are 
condensed and let the initial temperature of P, which has been 
observed by a thermometer placed in T, be t lf and that of steam t z . 

Once more allow the apparatus to dry and weigh the solid S. 
When steam is again passed into J with S in the pan, a greater 
amount of steam will be condensed on account of S. Let this 
now be W and suppose the initial temperature now is T v 

Previously the scale-pan condensed w grammes of steam and 

rose in temperature through the interval (t 2 — *,). In the 

second case it rises from Tj to t%. The amount of condensation 

w ' 

per degree rise of temperature is -r- — r~, hence the weight of 

(t 2 — *i) 

steam condensed in the second case is : -j- r-r x {t 2 — Tj). 



Denote the mass of the solid by m and the specific heat by s. 
The mass of steam condensed by the solid is : 

w(t t - TJ 


('• ~ h) 

Hence we have the equation : 

ms(t t - TO =fw - H' ! 




(*, - tj 

by means of which the value of s may be calculated from a 
knowledge of the value of L or, conversely, L may be determined 
if the specific heat of the solid is known. 

L denotes, as usual, the latent heat of steam. 

A correction to account for the differences in apparent weights 
of the solid in air and steam has been neglected. The temperature 
of the solid is supposed equal to that of the apparatus and 
surrounding air. An interval of certainly not less than twenty 
minutes is required to allow the solid to acquire this temperature. 

Bunsen's lee Calorimeter 

Description and Preparation for Use 

A diagram of the apparatus is shown in fig. no. The calori- 
meter is represented by ABCFED. It consists of a test tube, B, 

Fig. 1 10 
fitted into the glass jacket, A, which is drawn out at its base 
into the tube, CFE. This tube ends in a cup, E, closed by a 
cork, through which passes the narrow glass tube, D. 

A is filled partly with clean mercury and partly with distilled 
water containing no air. 


The surrounding jacket, J, is a calorimeter closed with the 
cork or wooden stopper, S, which supports the apparatus. 

In order to keep J and its contents at the freezing point, 
it is placed in a larger vessel, standing on non-conducting blocks 
and packed round with a mixture of ice and snow or with flaked 

By cooling the inner surface of B sufficiently a layer of ice 
may be formed round the outside, as indicated at I. 

On melting one gramme of ice the volume diminishes by 
•0907 c.c, so that if heat be added at B the amount may be 
determined by noting the change of volume as a result of the 
partial melting of I. The change of volume is observed by noting 
the movement of the end of the mercury column at D along 
the capillary tube. If this has been previously calibrated the 
change can be observed directly. 

In order to fill the apparatus, remove the capillary tube, D, 
and the stopper in E, and introduce into A sufficient distilled 
water to fill it to about half. Invert the apparatus with the 
open end of the test tube downwards and carefully boil the water, 
continuing until A is about one-third full. 

While this is proceeding, boil some distilled water in a large 
beaker, and towards the end of the evaporation of the water in 
A, place the end, E, well under the surface in the beaker. 

Cease boiling the water in the calorimeter, and allow more to 
flow over from the beaker. In this way the inside of the calo- 
rimeter and the tube, CFE, become filled. 

Clean mercury must now be passed in to lie below the water 
in A. 

Introduce it gradually from a pipette held under the surface of 
the water in E, allowing displaced water to overflow. Take care 
that no air bubbles are introduced with the mercury, particularly 
when it becomes necessary to tilt the apparatus to allow water 
to pass over the mercury in A towards the tube. Fill up with 
mercury to E, place the stopper in position, and by carefully 
adjusting it make the end of the thread coincide with any 
desired position along D. 

The apparatus should then be placed in a calorimeter containing 
water and ice to reduce the temperature as nearly as possible 
to zero. 

This will probably take an hbur at least, and the progress 
of the fall may be tested by placing a thermometer in B. 

When the temperature is about 2° C. introduce cooled ether 
into B. The ether may be cooled by placing it in a cooled test 
tube, and standing it in the calorimeter with the apparatus. 

Draw air through the ether and cause it to evaporate, con- 
tinuing until a cap of ice surrounds B. 



The solidification will cause D to move farther along the 
capillary, and enough ice should be formed to cause more 
expansion than is likely to be required in succeeding experiments 
with the apparatus. 

The evaporation of the ether may be brought about by some 
such device as that indicated in fig. 111. 




Fig. 111 

When sufficient ice has been formed the remaining ether is 
evaporated, and a current of air further drawn through to 
remove all traces of it. 

Now place the apparatus within the jacket, J, and stand it 
in the vessel, K, packing it round as described above with ice 
and snow. 

Leave this standing for an hour or two until a steady state 
has been reached, and the movement in the capillary tube is 
only slight and steady. 

It is not possible to maintain the end of the mercury thread 
quite steady with this arrangement, so that the slight motion 
must be accounted for in determining results from observations. 

The capillary tube may be calibrated by the method described 
on p. 41. 

Calibration of the Apparatus 

By placing warm water within B of known mass and tempera- 
ture, we may note the movement of D at various parts of the 
capillary tube for a known absorption of heat. 

In performing the experiment it will be sufficient to calibrate 
the tube for one particular region, and in using the instrument 
again a slight pressure on the stopper or a slight easing of it will 
drive the end of the thread into the calibrated strip. 

Heat up pure water to about 25 C, and transfer carefully 
to B. Allow the apparatus sufficient time to become steady, 
and note the displacement of the end of the mercury thread. 
Let this be /, and let the time be noted between the insertion of 
the water and the return to steady conditions. This will be 
denoted by t. 


In order to correct for the small creep of the thread, observe 
the rate of motion just before adding the warm water, and also 
just after the absorption of heat. 

All these measurements are to be made with a travelling 
microscope mounted conveniently opposite the capillary tube. 

If the rates of creep are respectively p and p 1 , then the average 
rate may be taken as \{f> + p 1 ) during the time t. 

Thus the displacement of the thread due to absorption of heat 
from the water is / — %(p + p x )t. We shall call this quantity L. 

If m is the mass of water added and its initial temperature, 
then a motion of L units corresponds to the absorption of md 

Determination of the Specific Heat of a Substance 

Let M grammes of the substance be heated to a temperature 1 , 
and let s denote the specific heat. 

The substance when placed in B will transfer to the ice M0 x s 

If the mercury thread moves a distance, L 1 , in the calibrated 

region, the heat absorption is -j— L 1 calories. 

Tu ,,., m0L l 

Thus M0 x s = — j — , 

md L 1 
S= Wi'L- 

The correction for Creep must again be applied by observing 
the motion just before and just after the insertion of the mass. 
L 1 is the corrected length. 

When a solid is put into B, a pad of cotton wool should be 
placed at the bottom of the tube to prevent breakage when it 
falls. In order to facilitate removal it is a good plan to tie a 
light thread round it. This will introduce only a slight error. 
During the absorption of heat, and generally while the apparatus 
is in use, the end of the tube, B, should be stopped with a plug 
of cotton wool. 

In the case of determining the specific heat of a liquid the 
experiment is almost exactly a repetition of the calibration. 

In order to dry the tube after liquid has been put in, a roll 
of clean blotting paper may be used. 

The Determination of the Density of Ice 

In this experiment we require to know the volume per unit 
length of the capillary tube. We have assumed that this has 
been previously calibrated. It thus remains to determine the 


shrinkage due to absorption of a definite quantity of heat. 

The experiment may be performed in conjunction with the 

calibration just described. 

Let L denote the latent heat of fusion of ice, and suppose 

warm water added to B imparts k calories. 

The amount of ice melted is -=- grammes. 

Let dv denote the shrinkage as measured by the movement 
of the mercury thread. 

Then -=- grammes of ice have become ■=- grammes of water 
L L, 

at o° C. 
Let d denote the density of water at this temperature. 

The volume of water is j-^ c.c, and the volume of the ice 

is thus : 

Gi + »)«* 

Hence the density of ice is : 



. grammes per c.c. 


The Latent Heat of Fusion of Ice 

It is assumed that the student is familiar with the principles 
of the determination of the latent heat of fusion and has carried 
out the experiment without making corrections for radiation. 
We are concerned in this description chiefly with an account 
of how this correction may be made. Care is taken, as in the 
determination of specific heat, to adjust the initial and final 
temperatures so that the room temperature is the mean of the 
two. *In addition, care must be taken that the calorimeter is 
not cooled down so low that the dew point is reached, otherwise 
there will be a deposit of dew on the apparatus, and a liberation 
of latent heat in consequence. 

We may make the correction for radiation as in the experiment 
on specific heat, but an alternative method will be described. 

Note the temperature when the ice is placed in the calorimeter, 
and at intervals of half a minute until it is melted, and finally 
at intervals of one or two minutes during which the calorimeter 
is absorbing heat by radiation. 

Let the temperatures observed in this second period be 

1 1> 1 2> I 3 • • • ■*• n+V 

Then the change of temperature due to radiation during these 
intervals will be : 

»e t = AftCTi + T 2 ) - t ] 
<50 2 = AftCT, + T 3 ) - t ] 

<50 o = A[£(T n +T B+1 ) -*„], 
by Newton's Law of Cooling, where A is a constant depending 
on the calorimeter but not on its temperature, and t is the 
temperature of the surroundings. 

Thus the total change of temperature A0 is given by : 


= A{ Tl + 2 Tw+1 +(T 2 +T 3 + ..+T n )-^ J 

But A0 = T B+1 - T 1( so that A can be calculated. 
In the first part of the experiment during the melting of the 
ice, let the observed temperatures be : t x , t 2 , . . . t n+1 . 
Then we have : 


AJ * 1 + J n+1 + (*, + t 3 + . . . + t n+1 )-nt Q }. 

Thus, since the observed minimum temperature is t n+1 , the 
corrected minimum is (t n+1 -j- At). 

In the above the <50's and dt's are to be treated algebraically, 
for some will be negative and some positive if the initial tempera- 
ture is adjusted in the manner described. 

Thus, if W is the water equivalent of the calorimeter and 
contents, and m the mass of ice melted, we have : 

wL + mt n+1 = W{*i - {t n+1 + At)\. 

The Latent Heat of Vaporization. (Berthelot's Apparatus) 

One form of this apparatus is depicted in fig. 112. Its essential 
feature is the condenser spiral, C, with the receptacle below. 
This is immersed in water in a calorimeter and the calorimeter 
shielded by packing it round with a non-conductor and placing 
it in a convenient vessel, or better still by standing it on non- 
conducting blocks in an empty larger calorimeter and enclosirig 
within the larger vessel (see fig. 112). 

The condenser is dried and weighed. It is placed in the 
water and allowed to stand until the temperature becomes steady 
as recorded by the thermometer, the uniformity of temperature 



throughout the calorimeter being procured by means of a 

A quantity of vapour is introduced into the condenser, and 
the spiral provides a large area of contact with a cold surface, 
so that liquefaction takes place and the liquid collects in the 
receptacle. The water is constantly stirred, and its temperature 
observed by means of the thermometer. 

Fig. 112 

When a suitable rise is obtained the supply of vapour is cut 
off and the thermometer watched until the maximum temperature 
is reached. In this interval, immediately after cutting off the 
supply, the end of the condenser outlet tube must be closed by 
a cork to avoid convection effects. 

The condenser is removed, dried, and weighed, and the mass 
condensed thus determined. Denote this by m, so that if L 
denotes the latent heat the supply of heat on condensation is 
mL calories. 

Let T 2 denote the temperature of vaporization, and T 2 the 
final temperature of the calorimeter and contents, while T 
denotes the initial temperature of the calorimeter. 

Let s denote the specific heat of the liquid, and W the water 
equivalent of the calorimeter, condenser, and remaining calori- 
meter contents. Then : W(T 2 — T ) =wL + m(Tj — T^s ; so 
that L may be determined. 

One of the weak points of this form of apparatus is the mode 
of introduction of the vapour. 


As the diagram shows, the liquid is vaporized over a small 
gas ring in the reservoir. It thus easily happens that the 
vapour gets superheated, and, in addition, it is difficult to 
shield the calorimeter effectively from the heat of the flame. 
The screen, S, is introduced to reduce this effect. 

It is preferable to use an electrical method of heating, and this 
is done in the more recent forms of the apparatus. 

Kahlenberg's heater is illustrated in fig. 113. The liquid is 
contained in the vacuum flask, H, and is heated by passing 
an electric current through the platinum wire, PP. 

For ordinary laboratory practice a test tube may take the 
place of H without introducing much difficulty in shielding the 

Fig. 113 

Correction for loss due to radiation in the calorimeter, C, must 
be made by one of the usual methods (p. 197). 

The Heat of Solution of a Salt 

When a salt is dissolved by a liquid the solution is accompanied 
by an absorption or liberation of heat. The amount varies 
with the proportion of the salt to the solvent, i.e. with the 
resulting concentration, and with the amount of salt dissolved. 

The number of calories absorbed or liberated when one gramme 
of a salt is dissolved in a certain amount of solvent is said to be 
the heat of solution for the particular concentration. 

The experiment described below is designed to measure this 



The apparatus necessary is illustrated in fig. 114. It consists 
of an outer protecting calorimeter of metal, C, which carries a 
cork through which the inner vessel, A, is supported. 

This vessel also carries a cork through which pass a thermo- 
meter, stirrer, and thin test tube, B, into which the salt may 
be placed. 

Fig. 114 

The vessel, A, and stirrer, S, are usually of glass since many 
solutions attack copper. There is uncertainty about the specific 
heat of the glass, so that the water equivalent of the vessel, A, 
and its contents should be determined in a separate experiment. 
For many purposes, however, we may assume the specific heat 
of the glass to be «i6. The pure solvent is placed inside A, and 
it should surround the lower part of the tube, B, which contains 
the salt. 

The apparatus is allowed to stand so that the salt may acquire 
the temperature of the solvent. Half an hour should be allowed 
for this, and during this interval the salt may be occasionally 
stirred by a clean dry glass rod. 

The weight of solvent is obtained in the usual way before 
the insertion of B. Let this be denoted by W. 

Suppose the water equivalent of A and its contents is w, and 
the specific heat of the solution s. This quantity must be 
determined later by the method of cooling (p. 199), or by any 
other convenient experiment. 

It will be necessary, also, to make a correction for radiation 
by one of the methods previously described (p. 197). 

Let the initial temperature of the salt and solvent be denoted 
by t and the final temperature, corrected for radiation, by t v 


and suppose that the weight of the salt dissolved is q. The 
heat of solution being Q, we have : 

qQ= {(W+q)s+w}(t-t ), 

for (W -f q) is the weight of the solution. 

The value of Q will be positive for the liberation of heat, and 
negative for absorption. 

In order to mix the salt and solvent, the glass rod used above 
for stirring should remain in the salt until just before mixing. 
At this instant observe the temperature, t , recorded on the 
thermometei^and push the rod through the bottom of the 
thin tub<T?T 

This tube must, of course, be clean and dry, and the salt will 
then fall into the solvent and be dissolved. The rod should 
then be removed, and care taken not to carry with it any of the 
solvent. The solution is assisted by the stirrer, S, and observa- 
tions of T taken every quarter or half minute in the initial 
stages, and later at longer intervals. 

A graph is drawn showing the variation of temperature with 
time from the instant of mixing, and this curve corrected for 
radiation as explained above, or we may use the non-graphical 

To find the Ratio of Specific Heats at Constant Pressure and Constant 
Volume for Air. (Clement and Desorme's Experiment) 

Apparatus and Experimental Details 

A glass reservoir, provided with a tap, T, giving a wide opening 
to the air is connected to an oil manometer, G, and to a pump; 

Fig. 115 

an ordinary bicycle pump is convenient. A small excess pressure 
is applied to A, the difference between it and the atmospheric 
pressure being measured by the manometer (fig. 115). 


In the first stage of the experiment the temperature under 
these conditions is allowed to become steady. 

In the second the tap, T, is opened and closed suddenly by 
giving one half-turn. 

By this means the pressure falls to that of the atmosphere 
in so short an interval that we may suppose there is no passage 
of heat to A during this expansion. 

The condition of the expansion is therefore adiabatic. 

Finally the temperature is allowed to return to that at the 
beginning of the experiment, during which process the pressure 
in A increases, though it does not recover its original value. 


Suppose that gas occupying the volume below the dotted line 
remains in the flask all the time. 

Denote its volume by V while that of the flask is.V. 

Let the initial pressure be P , and let that immediately after 
the adiabatic expansion be B. 

Then BV' = P V ', 

where r — ratio of specific heats, the value of which we require. 

The flask is open to the air so that B is the atmospheric pressure. 

It is in order that we may ensure a fall of pressure from P 
to B during a short interval that the tap is wide. 

In the final stage let the pressure become P when the tempera- 
ture has attained a steady value. 

We have now passed from volume, V , and pressure, P , by 
an isothermal process to volume, V, and pressure, P. 

/. PV = P V . 
/. P'V «= P 'V '. 

Hence by dividing this by BV f and P V f , we have : 




p r-x 

B . 

p * 


r — 




Let the difference in heights of the manometer be h initially, 
and h finally. 

Then if pressures be measured in terms of heights of the liquid 
columns : 

p = B + K, P = B + h, 


r = 

j—X (approx.), 

by expansion in logarithmic series and neglecting higher powers of 
~ and -4g — than the first. 

This is permissible since the values of h used are only a few 
centimetres. 5 cms. is a convenient value for h . 

Hence r — -, — 2-r- 

«o — A 

The result may also be obtained by another method. 

In fig. 116 let AC denote any curve relating the pressure and 

volume of a gas. The elasticity is denned to be the ratio : 



The stress will be measured by a slight change in pressure, 
and the strain by the corresponding slight change in volume 
per unit volume. Let us consider the volume, v, represented 
by BC. Let a change of pressure, dp, denoted by FA, bring 
about the change in volume denoted by CF. We shall record 
this by &o, but 6v = — CF on account of the diminution of 
volume on the addition of pressure. 


Fig. 116 


Thus, the elasticity, E, is measured by AF -7- — » for — 

J v v 

denotes the change of volume per unit volume. 

.-. E-— flt 


When we proceed to the limit and make the changes very 
small we have : 

E = — v-r-- 


For a gas the value of E depends on how the change is made, 
and we shall consider two cases, first the case of an isothermal, 
and then that of an adiabatic change. Let AC denote the 
adiabatic curve and AB the isothermal ; the former is steeper 
than the latter. Let E* denote the adiabatic and E«, the 
isothermal elasticity. 

From the formula for E we have : 

E = — v x slope of curve. 

Thus E^ == — v x slope of adiabatic, 

and E fl = — v x slope of isothermal. 

If we start at A with a particular volume, v, measured for each 
curve by P A, we have : 

E$ slope of adiabatic 
E^ — slope of isothermal 

The changes of pressure and volume in the experiment are 
small, so that the curves, AB and AC, are approximately straight, 
and the slopes can be measured by : 

p£ and gg respectively. 

•* E„~~AE 

Now, AF is the change in pressure during the adiabatic part 
of the expansion, viz., h , and AE is the change during the isother- 
mal part. In our case the atmospheric pressure is that at the 
end of the adiabatic expansion, i.e. that at C, and since the 
point, B, on the curve represents the final state, CB denotes 
the pressure, h. 

Thus AE = h Q - h. 

~,, Ea h n 



E, ~ 

' h — h 

For the adiabatic 
Differentiating we 


log^> + 
have : 

we have : 
pv r — constant, 
r log v = 0. 




dv ~ 

- — r. 


In our case p and v denote the values of the co-ordinates at A, 

id -£- i: 

axis of x. 

and j- is the tangent of inclination of the curve AC to the 


. Similarly, for the isothermal case we have : 
pv = constant, 

and £»_£ 

dv v 

p and v are the co-ordinates of A, but the slope is now for AB. 
Thus the ratio of the slopes is r. 

■ • - -* = r = h ° 

E«j h — h 

In carrying out the experiment make about six independent 
determinations, and increase the pressure cautiously so as not 
to expel oil from the manometer. 


Vapour Density. (Victor Meyer's Method) 

The vapour density of a substance can be found by measuring 
the volume of the vapour produced from a small quantity of the 
solid or liquid whose weight is known. In Victor Meyer's method 
this volume is found from the volume of air displaced by the 

The apparatus (fig. 117) consists of a vertical glass tube pro- 
vided with a bulb at the lower end, A, and a side tube, ST. The 
side tube dips under water in a beaker and a rubber cork closes 
the upper end of the tube. 

Fig. 117 

As it is necessary, on introducing the vapour, to allow a small 
bottle, D, to fall the length of the tube, it is advisable to place a 
little asbestos at the bottom of A to prevent breakage. 

The tube, A, is surrounded by a larger tube containing a liquid 
which boils at a higher temperature than the substance to be 
experimented on. 



In the case of the determination of the vapour density of 
ether, water may be used in the bath. 

The various parts of the inner tube should be kept at a constant 
temperature during the experiment, and in order to maintain 
this condition the outer tube is screened from draughts by 
surrounding it with a cylinder of asbestos or cardboard which 
fits it above the bulb. 

Before beginning the experiment the inner tube must be quite 
dry. If necessary it should be warmed over a Bunsen flame 
while a current of air is blown through it. 

The apparatus is set up as shown in the diagram, and heating 
is kept up until no more bubbles come out from the side tube. 

The substance is weighed and enclosed in the small bottle, D, 
and suspended close to the upper end of the inner tube. When 
everything is quite steady the bottle is allowed to fall ; vaporiza- 
tion takes place, and the cork of the bottle is blown out. Air 



Fig. 118 

passes over into the side tube and may be collected in the burette, 
B. It is better to collect it by the method illustrated in fig. 118, 
in which case the gas collected can be brought to atmospheric 
pressure by raising or lowering the burette. In the other case 
it is necessary to correct for the height of the water column, L. 

In order to cause no disturbance on introducing the substance 
to be vaporized a piece of thread or thin wire should be passed 
through the cork and be held by a stop-cock, E, which pinches a 
piece of tubing, F. The bottle is allowed to fall by opening E, 
and then closing it immediately. 

A better method is to use the apparatus shown at G. By 
turning the wire, H, through 180 the bottle will be caused to 
slip off the hook. 


The air collected at B is over water at a temperature, T, and 
will be saturated with water vapour at this temperature. Let B 
be the saturation pressure at this temperature. If v is the 
measured volume of air, and H the total pressure, then v the 
volume under normal conditions is given by 

H - B 273 

v = v x — 7 — x L ^f 

76 273 + T 

If w — weight of substance enclosed in D, the density of the 
vapour is : 

w __ w 76 273 4- T 
v ~ ~v H -B 273 

1 c.c. of hydrogen at o° and 76 cm. pressure, weighs -0000900 
grammes, and its molecular weight is 2. 
Hence the molecular weight of the substance examined is : 

2 w 76 ^ 273 + T 

•00009 v H — B 273 

At the conclusion of the experiment remove the stopper from 
the end of the inner tube to prevent any sucking back of water 
from B into the bulb, A, as the apparatus cools. It is important 
to cause the bulb, A, to be heated by the steam from the bath, 
and it should be adjusted to prevent actual contact with the 
water in the bath, and to be out of reach of splashes when boiling 
takes place. 

Vapour Density. (Dumas's Method) 

A large flask is cleaned and dried, and fitted with a cork provided 
with a bent piece of glass tubing drawn to a fine point so that 
it can be easily sealed by the application of a Bunsen flame 
(fig. 119). 

The flask is suspended from the arm of a balance and weighed. 
Let the observed weight be W lt and let w x denote the weight of 
air displaced by the wall of the flask, while w 2 denotes the weight 
of air displaced by the closed flask, so that w 2 — w x denotes the 
weight of air within it. This will be denoted by w a . Hence 
if W denote the real weight of the flask : 

W 1 = W-ze' 1 (i) 

Now introduce a small quantity (5-10 c.c.) of the liquid to 
be vaporized into the flask. The actual quantity will depend 
on the size of the flask and will be discovered by trial. The 
amount given is of the right order when a flask of capacity about 
a half litre is employed to find the vapour density of chloroform. 
This liquid is very suitable in the case of a laboratory exercise, 
since it has a high vapour density. The flask is placed within 


an enclosure, J, over a sand bath, with the tube, T, projecting 
through the lid. 

Heating is continued until the liquid is vaporized and no more 
issues from T. This may be tested by placing a polished surface 
near the end of T. It will become dimmed if vapour is still 
coming out. 

When the steady state is reached, T is sealed off. 

Fig. 119 

Let this happen at a temperature, t°, measured by means of a 
thermometer hanging close to A, within the enclosure, J, and let 
the volume of the vapour be V t , and density ? t . Allow the flask 
to cool, and weigh. Break off the end of T under water. The 
flask will fill with water, the space occupied by the condensed 
vapour becoming negligible under the new conditions. 

Preserve the broken pieces from T, and after drying the outside 
of the tube re-weigh the flask and water, noting the temperature 
of the water, t °. 

Let W 2 denote the weights in the scale-pan when the flask 
and vapour are weighed, and let w v denote the weight of the 

Then W 2 = W + w v - w 2 ; 

.-. W 8 — Wj = w v — {w t — Wj) by equation (1) 


Let V t0 denote the volume of the flask obtained from the 
weight of water it contains at t °. Then if p denote the coefficient 
of cubical expansion of the glass, which may be taken as '0000232, 

V t = V i0 {1 + ?(<-«„)}, 

If the barometric pressure be P, we can reduce p t to normal 
conditions by the formula : 

(273 + *) , 76 
273 P 

(273+J) . 76 w % 

H =~m T' Pt 

2 73 P V #0 {1 + fi(t-t Q )}' 

w a denotes the weight of air filling the flask at the temperature 
of the air within the balance. Let this be t 1Q C. 
Then if d a denote the density of air under these circumstances, 
w. = d.V* = da V f() {1 + p (V- - t Q )\ 

-^•^.^f^r-v^i + p^-^)}, 

d — -001293 gm. / c.c. 

From these two formulae, since w v = w a + W 2 — W 1( we may 
now calculate p . 

One of the difficulties of the experiment is to drive out all the 
air from the flask and replace it by vapour. It frequently 
happens that on attempting to fill the flask with water some air 
is left behind. More of the liquid is required for driving off the 
air in this case. 

If the volume of air left over is small, we may apply a correc- 
tion. At the temperature t , of the water, let the volume of air 
be v. This may be determined by filling up the flask with water 
from a measuring flask. 

The total pressure to which the mixture of vapour and air is 
subjected is the sum of the partial pressures, P 1 and p\ of the 
vapour and air respectively. The weight of the vapour is now 
w v \ obtained by subtracting the weight of air of volume, v, from 
w v , determined as above : 
wj> = p% 

V <0 denotes as before the total volume of the flask, i.e. the 
volume of the water after the bubble of air has been replaced. 

We can find P 1 by remembering that the air of volume, v, and 
at atmospheric pressure occupied a volume, V t , under the partial 
pressure, p 1 , the temperatures being respectively t ° and t° C. 


aP = V t p l 

273 +* 273 +r 

p 1 = P -P 1 ; 
.-. P x V t = PV t - 273 1 1 • Pv, 

* * 273 + 1 

273 + * 76 I 273 + 1 J 

or p = 

w J 

273. P 

L 273 + 1 273 + *oJ 

76 L 273 + 1 

The term, — » will be small if the experiment is to be 

273 + 1 y 

successful at all, so that it will only be necessary to change the 
value, w v , to the actual weight of vapour, to, 1 . 

It should also be remarked that the volume, v, is a mixture of 
air and water vapour, whereas it has been assumed to consist of 
air only. We shall not, however, further consider this correction, 
which is small and affects a term which must be already small. 
The discussion shows how the error affects our formula, and it 
would be sufficient to measure the volume of the bubble, multiply 
by the density, -001293 gm./c.c, and subtract from w v . 

After a few attempts the bubble will usually be sufficiently 
small to be neglected altogether. 

Conductivity of a Copper Bar 

The apparatus consists of a bar of copper, CC (fig. 120), with 
two holes bored well into it to carry thermometers, T x and T 2 , 
mercury being placed in the holes to ensure good thermal con- 
tact. At the ends of the bar are two metal boxes, through one 
of which, A, is passed steam, and through the other, B, a steady 
stream of water from a constant pressure head. 

The shelves, LL, within B, serve to prevent any flow of water 
straight from inlet to outlet. The temperature of the water is 
taken just before it enters and leaves B by the thermometers, 
T 8 and T 4 . The apparatus is allowed to attain a steady state, 
when all the thermometers will record steady temperatures. 

It is usual to pack loosely round CC and the boxes some 
cotton wool, the whole being enclosed in a felt-lined wooden box, 
through which T t and T a project. T 3 and T 4 are kept as close 


as possible to the box, and it is a good plan to wrap the T-pieces 
loosely with wool. 

In this way the heat from A is transmitted by conduction 
to B, and the heat passing across in any time, t, is noted by 
collecting water as it leaves B and weighing it. If the mass 
collected is m the rate of transmission of heat is 

*» (T 4 - T 3 ) 

t '• 

T. T, T, 

f 1 

j-L— 3 
B Tc 

-L— } 

C A 

Fig. 120 

T 3 and T 4 denote the initial and final temperatures of the 
water. Since this quantity of heat is transmitted from T x to T a 

—a distance, d, say, the amount is k • (T 1 — T a ) • j, where A is 

the area of section of the bar, and k the conductivity 
A is measured by finding the diameter of the bar if it is cylindrical, 
or by measuring its breadth and height if rectangular. 
We have, therefore : 

k (Tj — T a ) • j = m 

(T 4 - T 8 ) 

In another form of apparatus the cold water is passed through 
a metal spiral wound round the end instead of through the metal 
box. Good thermal contact is made between the spiral and bar,^ 
and the temperature of the water measured at entrance and exit 
as before. 

Thermal Conductivity of Rubber Tubing 

The apparatus required consists of a length of rubber tubing, B, 
a copper heater, A, for producing steam, a calorimeter, C, 
thermometer, T, and a measuring glass, as illustrated in fig 121. 

The method of procedure is as follows : 

A quantity of water is introduced into the calorimeter, C 
and weighed. 


Steam is passed through B until a rise of temperature of the 
calorimeter and water of from io° to 20 ° C. has occurred. The 
initial and final temperatures are noted and also the time of 
passage of the steam. The tube is removed and the length that 
has been immersed is noted — I (say). Two pieces of cotton 
should be tied round the rubber at the points where it enters 
and leaves the water. 

Fig. 121 

Let the initial temperature be T and the final T 1 . If there 
had been no loss of heat by radiation the final temperature would 
have been some other value, T 1 + A T, and it is necessary to make 
the correction AT. 

In order to do this, observe temperatures, at intervals of half a 
minute or some convenient period, beginning at T and ending 
at T 1 , during the passage of the steam. 

These will be denoted by t v t a , ... t n . x . 

If the temperature of the room be t we have for the change 
of temperature due to heat radiated or absorbed, as the case may 
be, according to Newton's Law of Cooling : 

T +t x 


dd 2 = C J — t [, 


dd n = C 

. . \ ... c 


where 3d denotes the change in any interval, the suffix denoting 
which interval. 

AT = S.50 = C | T ^ T1 - w* + (*i + *i + • • • + *«-i)|. 

C is a constant depending on the calorimeter and contents, 
and must be determined if M is to be calculated. 

When the final temperature, T 1 , has been attained, cut off the 
supply of steam and allow the calorimeter to cool, observing 
temperatures, T^ T 2 , . . . T B , at equal intervals. 

Then as before if 8t denotes the loss of temperature in each 
interval we have : 

S# = C |3i±I-« - (»-i) t + (T a +T 3 + . . . + T.-0 

But S# = T x — T n , so that C can be calculated and may be 
employed in the above case. 

The ranges of temperature in the two cases should be as 
nearly as possible the same. 

If M denote the water equivalent of the calorimeter and con- 
tents the heat transmitted through the tubing is 


We can connect this with the conductivity, k, in the following 
way : 

Let the outer and inner radii of the tubing be r x and r 2 , and con- 
sider a portion of unit length of the tube between radii, r and 
r + 8r, at which the temperature is t. 

The rate of change of temperature at the distance, r, is — -j-. 

The negative sign expresses the fact that the temperature 
diminishes as r increases. 

Thus if Q denote the quantity of heat transmitted per sec. per 
unit length, i.e. across an area, 2nr, 

r» t, dt 

J 2-ak J r 



where the integration is to be taken between the limits, r x 
and r 2 for r, and between the inner and outer temperatures of 
the tubing for t. 

Let the steam temperature be t, the outer temperature is 
taken as the mean of T and T 1 , 

/. ,- i<T+T') = 2 -filog/J. 

But Q = y (T 1 + AT - T), 

. . M T> + AT - T , r, 

■■ * - 55 x 2 ' 3 ° 3 x utt ' lo e» f, 


(Change being made to logarithms to base io.) 

The value of r x may be determined by means of a screw gauge, 
and in order to find r 2 , place a length of the tube of 5 to 10 cms. 
in water in a measuring glass and note the volume, v, displaced. 

Then if L denote the length of tubing : 

v = tcL (rS - r 2 *). 

All the quantities except r 2 are known, so that this value can 
be determined. 

Another way of determining the radii is to cut the tube clean, 
normal to its length and use it as a rubber stamp, pressing it 
lightly on a clean sheet of paper. 

The impress of the outer and inner circumferences will be 
distinct and the diameters may be measured by means of a 
travelling microscope. 

The Conductivity of Glass 

The conductivity of glass in the form of a tube may be found 
by the method described in the last experiment. A different 
arrangement of apparatus is required, but the theory is identical 
in both cases. 

Steam is passed through a jacket, J, round the tube, B. Within 
B a stream of water is caused to flow from a supply which provides 
a constant head of pressure. 

Within B is a spira> made of cord or rubber so that as it pro- 
gresses up the tube the water is caused to traverse it spirally. 
This is important as the temperature at any cross-section of B 


must be the same throughout the section. The rate of flow is 
adjusted to cause a difference of temperature of about 20 be- 
tween T and T 1 . The thermometers are enclosed in T-pieces 
as near as possible to the ends of B, the T-pieces being covered 
with felt or cotton wool to prevent loss of heat by radiation 
before the temperature is taken. 

Fig. 122 

In order to measure Q, water is collected on exit for a measured 

Conductivity ol Cardboard by the Method of Lees and Chorlton 

The apparatus consists of a retort stand provided with a clamp 
(fig. 123) and metal ring, AB, from which hangs a cylindrical 
slab of copper or brass, DE, of diameter about 12-5 cms. On 
this rests a hollow cylinder, C, of the same diameter, provided 
with inlet and outlet tubes, G and H, through which steam may 
be Dassed. 

Towards the base of C and into DE holes are bored so that the 
thermometers, T x and T t , may be inserted and the temperatures 


The two cylinders are nickel-plated in order to produce a 
surface of uniform emissive power. 

Suppose a thin slab of material of the same diameter as the 
cylinders is placed between them, and let the loss of heat that 
is radiated from the edge of the slab be small enough to be 
neglected in the calculation. Then all the heat transmitted 
across the slab is radiated from DE during the steady state. 

Let A denote the area of cross-section of the slab and d its 
thickness. Let the thermal capacity of DE be denoted by W, 
and let the thermometers, T x and T 8 , record temperatures, T x 


and T 2 in the steady state, while the temperature of the surround- 
ing air is T . The heat transmitted through the slab per sec. is : 





where k is its thermal conductivity. 

A B 





'T 2 


12 I 

Fig. 123 

The heat radiated per second from DE is 

C • (T 2 - T ), 
where C is a constant. 



A = C (T 2 - T ). 

We may readily use the apparatus to give comparative results. 

If two sheets be cut of the same diameter as DE, one of glass 
and the other of cardboard, about 1 mm. thick, and if they be 
inserted between the cylinders we may observe two sets of 
temperatures, one for each. Let the letters without dashes be 
used to describe the experiment with glass while those with 
dashes correspond to cardboard. 

In the latter case we have : 


T/ - T 5 

•A = C(T 2 '-T / ). 


On dividing we have : 

k' _ T 2 ' - T ' d' 
k T 2 — T n d 

T t -T s 

T 2 - T„ d T/ - T 2 ' ' 

d and d 1 are measured by means of a screw gauge. 

In order to find T place the thermometer below a sheet of 
cardboard, FF, to protect it from direct radiation from DE, but 
it should be placed directly below DE, so as to give the tempera- 
ture of the air which rises upwards to DE. 

An absolute value may be found for k by determining the rate 
of fall of temperature of DE at the temperature, T 2 . 

This may be found by removing C and allowing a Bunsen 
flame to play on DE until T 2 registers a temperature about io° 
above that recorded during the steady state. 

Observe the temperature recorded by T 2 as the slab cools to 
about io° below that recorded during the steady state. The 
observations should be made every half-minute, or more fre- 
quently if the change is rapid, and a graph drawn relating the 
time and temperature. 

From the graph determine the value of the rate of change at 
various temperatures. 

Plot a second graph with the rates, -^-as ordinates and the 

corresponding temperatures, T, as abscissae (fig. 124). From 
this determine the particular rate for the temperature, T a . 

Denote this 


Weigh the slab and determine its water equivalent, using for 
the specific heat of copper the value -094, or of brass the value -09. 
Denote this by W. 

Hence the loss of radiation is also expressed by W( — J 

and we have the equation : 

by means of which k is determined. 


Make a table of the observations during cooling as follows : 





RATE— . 








Forbes's Method of Determining the Conductivity of a Metal Bar 

This experiment consists of two parts, in the first of which a 
metal bar is heated at one end until the steady state is reached. 

In the second part a bar of the same material and cross-section, 
but shorter, is allowed to cool under similar external conditions. 

In order to maintain constant external conditions the experi- 
ment should be performed in a part of the laboratory sheltered 
from draughts. 




The bar (fig. 125) usually has one end curved and dipping into 
a convenient molten metal contained in a vessel on the other 
side of a screen, which protects the bar from direct radiation 
from the source of heat. 

The metal may conveniently be molten lead or solder. The 
bar is provided with a series of holes which lie regularly along 
its length into which thermometers fit. 


If these are small and contain mercury or the molten 
liquid, it is found that the process of conduction is not 
disturbed by their presence. The holes near the hot end should 
contain the molten liquid and the remainder mercury. 

By observation of the temperatures at various distances along 
the bar measured from the scifeen, the temperature slope may 
be found. This is best done by plotting a curve (fig. 126) and 

calculating the slope, -=-, from the inclination of the tangents 

at various points. 



Fig. 126 

A thermo-junction may be used alternatively to find the 
temperatures by dipping one junction successively into the holes. 
For the calibration of the junction the reader is referred to p. 545. 

The importance of maintaining steady external conditions 
will be appreciated in this part of the experiment, and great 
care will be required to maintain the whole length of the bar 
simultaneously steady. 

The bar is assumed to be sufficiently long that its end is at the 
external temperature, O . A convenient length is one metre and 
its section may be about 2 cms. square. 

In the steady state all the heat passing a section, B, escapes 
from the surface between B and the end. 

The rate of flow at B is : 


where k is the conductivity and A the area of section. 

This rate of flow is calculated from the second part of the 
experiment by determining the rate of cooling of the portion of 
the bar beyond B. The equation thus obtained serves to find k. 

The second bar may be conveniently 10 cms. long. It is heated 
to the temperature of the molten metal, but so that its surface is 
not damaged and remains similar to that of the first bar. To 
do this and at the same time to prevent sudden cooling of the 
molten mass, the rod is wrapped in several layers of paper and 
completely immersed. 


It is provided also with a hole to carry a thermometer and is 
suspended under the same conditions as prevailed round AB, 
and its temperature observed at successive times so as to include 
in the range those values which prevailed along the bar. 

A curve is drawn showing the relation between the temperature, 
6, and the time, t, for the short bar. 

This bar is similar to the long bar and cools under the same 

conditions, so that the rate of cooling for both is the same. 

From the curve we can deduce the rate of cooling, — , in the usual 


We require for the purpose of the calculation the rate of cooling 

for points along the bar in the first part of the experiment. We 

must correlate the values of x and —=- , x denoting distances 

measured from the screen. 

This may be done from the curves. 

Take a series of values of x from the curve illustrated in 
fig. 126 and observe the corresponding value of 0. From the 

curve relating and t take the values -=- for these particular 

values of 0. 


We then have the corresponding values x and -=- 


Plot these on a curve as illustrated in fig. 127. 

Fig. 127 


This curve will cut the axis at a point, C, where -^ vanishes, 


or where the temperature of the bar is equal to that of the sur- 
rounding air. 

If 5 is the specific heat and p the density, the rate of loss of heat 
between two points separated by distance dx is 

AdXpS -57- 

dt • 
The total loss per second between B and the end of the bar 
is thus : 




Aps 5— ax 
b « 

* Ap Vb -s- 

= Aps x shaded area of fig. 127, 
graph measures the distance from the hot 


where OP on 
end up to B. 

This area may be found by the planimeter or calculated if 
carefully drawn on squared paper. 

•■• * A GD B =Apsxs - 

S = shaded area and ( •-=- ) denotes the temperature slope at 
the point B. 

■••*•(&- * 

We can thus determine k, the values of p and s being given as 
constants of the apparatus or determined in the usual way. 

These values may be taken to be 8-93 gm. per c.c. and -094 
respectively when the bar is of copper. 

The Determination of the Conductivity of a Bar of Metal by Angstrom's 

In this method heat is supplied to a long bar by alternately 
heating and cooling it at one region in regular periods. 

In this way the temperatures at points along the bar fluctuate 
periodically, and on account of surface radiation the temperature 

Fig. 128 

amplitudes diminish as the distance from the region of supply 
increases, while the maximum and minimum values occur at 
later timefe with the increasing distance. 

The bar must be chosen so that it is sufficiently long to allow 


us to neglect the effect of the terminal faces. The fluctuations 
should die away at a short distance from the cooler end, which 
thus has the same temperature as the air surrounding it. 

When the heating is continued long enough the periods develop 
themselves completely, in which case the mean temperature at 
any point of the bar preserves some constant value. 

We first consider the theory of the experiment. 

Let us consider a bar, AB (fig. 128), with one end, A, exposed 
for a definite time, T x , to a current of steam and for a succeeding 
time, T 2 , exposed to a current of cold water. Suppose that 
this process is continued regularly until the steady fluctuations 
throughout the bar are developed. These fluctuations will have 
a complete period of (T x + T 2 ) which we may denote by T, and 
for the sake of convenience we shall write 

Let the conductivity of the bar be, k, p its density, s its specific 
heat, A its cross-section, P its perimeter, and its temperature 
at a point, P, distant x from a convenient origin. We shall 
choose this to be at A. 

Consider a point, P 1 , distant dx from P, as in the diagram, and 
let its temperature be (0 + 66). The heat flowing into the 
element, PP 1 , of the bar across a plane at P drawn normally to 
its length is equal to : 

, P — P i 

dx A P ersec -> 

where the subscripts denote the points at which is measured, 
i.e. the heat flow amounts to : 

- k '~Sx~' A 

-- k 'Tx' A = F ^ s ^' 

in the limit when dx is made infinitesimal. 

Thus this expression denotes the flow from left to right at the 
point P, of the bar. Since depends on x, in the general case 
this flow will also depend on x, i.e. F depends on x. 

Again consider the element PP 1 . We have calculated the 
flow, F, into it at P, and the flow out at P 1 will be 

F + — • dx. 

Hence the total flow of heat into PP 1 
^ _d¥ 



J, d * 6 A * 

= ft ' -j— z • A • 6x per sec. 


This heat is used up, partly in warming up the part of the bar 
concerned, and partly in radiation from the surface. 

If the temperature is changing at the rate, -j-, t denoting the 

time, the first part amounts to : 

and the second to : 

h • P • 6x (0 - O ). 
h denotes a constant for the surface and O is the external 

We shall, however, suppose that is measured in degrees 
above the surrounding temperature, so that we may write the 
last expression simply : 


We, therefore, arrive at the equation : 

kAdx ^ = h • P -dx • 4- A -gsdx -=? 
ax* at 


2 -*&-=•■ <*> 

where K =~ , H = -^- (3) 

ps Aps 

Any function which satisfies certain conditions can be expanded 
as a Fourier Series. 

Students of Physics should make themselves acquainted with 
the Fourier Analysis and they may be referred to Carslaw's work 
on this subject. For the present purpose we need nothing more 
than the statement that the expansion is possible. 

Fourier's Series is the following : 

f(t) = A + A x cos pt + A 2 cos 2pt + A 3 cos 3pt + . . . ) , * 
+ B x sin pt + B 2 sin 2pt + B 3 sin ipt + . . . J w 

Here f(t) is any function existing over some interval from 
t = a to t = b and satisfying certain conditions with regard to 
continuity, etc., which we need not enter into here. We merely 
mention that the series is applicable to the function we shall use 
in our experiment. 

With regard to the values of the A's and B's the rule is 


A B =^ / /(«) cos -7 — — u du; 

b — aj a b — a 

B„ = T -/ t(u) sin T w. du; 

b — a J t JK ' b — a 

n has the integral values, i, 2, 3, etc. 


The value of p in the above series is in this general case r 

When f(t) is periodic and has the complete period, T, we take the 
limits, a and b, at the ends of this period and write a = o, b = T, 
so that the series represents f{t) from t = o to t = T, and on 
account of its periodic character represents it also from T to 2T, 
etc. In this case : 

. 2tc 
P = x ' 

I I /* T 

Aj L-J f{u)du; 

2 /* T 
A B = =r / /(w). cos npu du ; 

2 f T 
B„ = — / /(w) sin w^)w • ^w. 


The periodic function with which we are concerned is that 
which expresses the fact that from t = o to t = Tj the temperature 
has some value, $ lt and from t = Tj to t = T 2 it has the value 2 . 
Such a function can be represented by a Fourier Series and we 
shall assume that the particular expression is (4), with coefficients 
calculated according to the rule. These, however, are not 
required for the experiment. 

It is shorter and convenient for our purpose to write (4) in the 
form : 

a Q + a x sin {pt + r x) + « 2 sin (2pt + r 2 ) + (5) 

And let us suppose that this is the expansion which expresses 
the temperature at A. 

Now the temperature, B v at the point whose distance is x t 
from A is a fluctuating function and so is that for any other 
point, %. 

We may therefore write for 6 lt some such value as (5), viz. 
B x = C ' + C/ sin (pt + a/) + C 2 ' sin { 2 pt + <V) + • • • • (6) 
and for the point, x % 

6 2 = C ' + C ± ' sin {j>t + d{) + C/ sin ( 2 pt + «■*) + . . (7) 

The quantities C ' C/ . . . , <V 6 2 ' . . . , will differ from 
C " Cx" . . . , 5/ 5/ . . . , since they correspond to different 
values of x. 



We have to solve the equation (2), and we have a clue from the 
experimental observations of diminishing amplitudes and lagging 
maxima and minima. Moreover, the value of is known at the 
point A where x is zero. This value is given by (5). 

The solution is in fact : 

6 = a,?-*** + a 1 e- a ^ sin (pt + p,* + r x ) + 
a#-** x sin {2pt + M + r 2 ) + etc (8) 

Where : a » = ? K (a n * - p n *) = H, \ {g) 

2Ka n p tt = -np J 

When x = o this expression reduces to that of equation (5), 
so that the solution satisfies the end condition. 

The reader maj' verify that the solution satisfies equation (2) 
by substitution. He will observe that the coefficients of the 
separate sine and cosine terms and the terms independent of 
trigonometrical functions all vanish if (9) holds. 

Now if we consider the two points, x x and x 2 , at a distance, I, 
apart, by equations (6), (7) and (9) we have : 

C/ = a x e~ a **\ C x " = a x e-"***, d x " - <V = p^. 

For S x = $ x x x -f r x and d x = $ x x 2 + r x 

Hence «. x l = log ~i and a^ = ' ^ x log ^V 

But by the last of the conditions (9) 


K = ■?— = TT-s «. (10) 

and in the same way : 

K = pr-, (11) 

T(d n ' -d n ")\og^ fl 

In this formula we note that the constant of radiation, h, does 
not appear and we are not troubled with the difficulties always 
associated with it. The conductivity, k, is given by Kps so that 
if our experiment is performed carefully and as a consequence 
the quantities, C n and 6 n , accurately known, we have to rely on 
the accuracy of the knowledge of the density and specific heat. 
Both these are accurately known. This is the reason of the 
importance of Angstrom's Method. The student is recommended 
to refer to a translation of the original paper in the " Philosophical 
Magazine," series 4, vol. 25, p. 130, 1863. 

From (11), we observe that each coefficient, C n , gives rise to a 
value of K. As the carrying out of the experiment will show, the 
coefficient, C x ; gives the most reliable result but the value corre- 


sponding to C 2 should also be worked out. It remains to describe 
the experiment and to show how to find the C's and <5's. 

The bar, which may be of copper, iron or brass, has one end, A, 
inserted in a chamber so arranged that steam and cold water 
may be passed in alternately as described above. The total 
period is to be measured and care exercised to reproduce the 
conditions exactly in each successive period. 

Steam from a conveniently large tin flask and water from the 
tap will set up the right conditions. 

T\me or the f>o»r>r« =c , 8- oc^ 


&, So 3G 

Fig. 129, 

The steady temperature fluctuations will be the more easily 
attained if the bar is sheltered to avoid draughts in its neighbour- 
hood and consequent troublesome convection effects. The 
temperature at points on the bar should be observed by means 
of a thermo-junction placed in a small cavity in the bar. Two 
such cavities will be required and should contain mercury, or 
it will be sufficient to hold the junction in contact with the bar 
at the points concerned. The thermo-junction must be calibrated 
in the usual way (see p. 545). 

When this is done read the temperature at convenient intervals 
by means of the galvanometer in the circuit and obtain as many 
readings as possible throughout two complete periods, or more. 
Plot a curve for 6 against the time very carefully on squared 
paper as illustrated in fig. 129. Suppose that this refers to the 
point, x v Repeat the process for x v 


We shall determine the necessary quantities from the graphs. 

In order to see how to use the graphs multiply the equation 

(6) by sin pt and integrate both sides over a complete period 

from any arbitrary time, t=t x to t=t x +T t where, of course T= — 

We thus have : 

0! sin pt dt = Co' / sin pt dt + C/f sin (pt + <5/) sin pt dt. 

+ C 2 ' f sin (2pt + <V) sin pt dt + etc. (12) 

In doing this it is well to remark that we are integrating the 
series on the right-hand side term by term, adding all the in- 
tegrated terms and equating to the integrated function on the 
left. This process is not always legitimate, but on account of 
the properties of the particular series we may apply it in the 
present case. 

On performing the integration the only term on the right-hand 
side which does not vanish is the second, and this has the value : 

^TC/cos <V- 
Denote the integral on the left by S,. 
Then S, = £ TC/ cos d x '. 

In the same way if S« denote the value of 

^,+ T 

B x cos pt dt, 

J t x 

we find 

S^iTC/sin V- 
We can determine C x and 6 X from these two equations, for 

tan V=|^ .(13) 

and C t '» = -^-(S,* + V) (14) 

In the same way if we multiply the series by cos 2pt and 
sin 2,pt and integrate we find : 

X cos 2pt dt = £ TC a ' sin <5 2 ', 

and as before we can express C a ' and «5 2 ' in terms of the quantities, 
S ac and Sg,, the meaning of the latter being of course 


0j sin 2pt dt sm $ TC a ' cos <5 3 '. 

Jto. j* . 


This suggests a graphical method of determining the necessary 

Measure as many ordinates of the curve, fig. 129, as is con- 
veniently possible and multiply each by sin pt or sin -=r . 

T is the complete period of the periodic heating and t is the 
value of the time appropriate to each ordinate. 

Plot a new curve with these new quantities as ordinates and 
times as abscissae and thus obtain fig. 130, 

Cutue sWxjoinQ \V»e uatia\tor> of 0sinSTTP ,jj\tVt 
\ke rime -fa* the \x>\r\X X.| T 

Fig. 130 

Draw any ordinate A X B X for some arbitrary time, t v and 
construct C X D X the ordinate at one complete period later. 

Measure the area between these ordinates, the curve and time 
axis carefully with a planimeter. 

This gives S,. 


Go through the similar process to find S c . 

Multiply Bx by sin 2pt and cos 2pt to find S 8 , and S 2C . All 
this has to be repeated for the second point, x 2 , and we then have 
all the data necessary to deduce k from formulae (10) and (11). 

C?otoe sWou/mq tW uwriatlons of ■& coe -=^r UJ»m 
the Hme at Fhe ^pomf 0C t 

Fig. 131 

Another way of making the calculation is to make two ex- 
periments with different periods, T' and T". If <x n ' and <x„" are 
the values of a n in these cases respectively we find from (9) the 
result : 


|t/2 „ /i 
'\ „ /2 

V 2 «» 2 . 

2a « a n * a n " — a, 
a n ' and a n " are derived as before, and 



It is to be noted that we do not in this case require the values d. 

The density and specific heat may be taken from tables or be 
measured by any of the usual methods. 

In drawing the curves it is a good plan to have two complete 
periods shown to give a means of testing the accuracy of the 
areas, S, and S c . 

Always test the end of the bar remote from the heat supply 
to determine if it remains at the air temperature. It will suffice 
to place a thermometer close to this end. A longer bar must be 
used if the fluctuations continue right to the end. 

Angstrom used a square bar of side 2-375 cms., and the length, 
I, between the two points, x x and x 2 , was 5 cms. 

It has been found, however, that a cylindrical bar of diameter 
from 1 to 2 cms., and of length about 60 cms., will suffice, 
with such a range of temperature as described above. In the 
original experiment Angstrom heated his bar at the central 
region and, of course, had similar conditions on both sides 



Determination of the Radiation Constant. ("Phil. Hag.'*, ser. 6, 
1905, p. 270) 

By Stefan's Law the total radiation from a black body is propor- 
tional to the fourth power of the absolute temperature, or 

R = oT*. 
It is the object of this experiment to determine a. 


The diagram (fig. 132) illustrates the arrangement of apparatus 
which consists of a blackened hollow metal hemisphere, B, about 
ten inches in diameter, fitted into a wooden box, W, lined with tin. 

Fig. 132 

This fits on to a table, of which the top, DE, is shown, containing 
a small hole at S, which lies at the centre of the hemisphere. 
B is heated to a uniform temperature measured by the thermo- 
meters, T x and T t , by passing steam through the box above 
the hemisphere. The black surface of B is the radiator, and 
the heat is received by a small disc of silver placed at S, and 
blacked on the upper surface to prevent reflection. It is better 
to fit the disc in a vulcanite frame rather than to allow it to 
touch the table directly. 

From the disc are lead away two wires, one of constantan, 
and the other of silver, to a galvanometer and second junction. 



S is thus one of the junctions of a thermo-electric couple, 
the other is placed in a tube containing oil standing in a calori- 
meter, C, containing water or ice. 

The junctions to the galvanometer are kept in a canister, A, 
packed with cotton wool to prevent any electrical effect due to 
difference of temperature at these junctions. 

It may be further necessary to include a resistance in the 
circuit to keep the deflection on the scale if the galvanometer 
is too sensitive. 

A rise of temperature due to absorption by the disc is thus 
recorded on the galvanometer. 


Let Rj = radiation absorbed by the silver disc per unit area 
per second, and R that emitted. Let the temperature of the 
radiator be T 1} and of the disc T. 

If the whole enclosure, the disc included, had temperature T lf 
there would be equilibrium, and the disc would both emit and 
absorb R x in unit time. The energy absorbed would arise, of 
course, from B. This same energy falls on the disc when at the 
lower temperature and is absorbed, but the energy emitted is 
now R. Thus the gain of energy per sec. 

= (R t - R)A, 
where A denotes the area of the disc. 


Let m denote the mass of the disc, s its specific heat, and -57 


its rate of change of temperature. 

Then we have : 

ws ¥ = ^f-- A = X (Tl4 " T4) ' 

or Jms _ dT 

AfV-T*) dt* 
where J = Joule's equivalent (4*2 X io 7 ergs per calorie). 

All the quantities on the right are measured, and hence a is 

Experimental Details 

It is first necessary to ascertain the" relation between the 
readings of the galvanometer scale and the difference in tempera- 
ture between the two junctions. 

In order to make this comparison the disc is surrounded with 
cotton wool, and the cold radiator placed above it. 

The calorimeter, C, is then heated and the difference in 
temperature between it and the disc recorded on a graph against 
the readings of the deflection. 


We thus obtain : difference in temperature per scale division = 

BC' J 

By measuring the temperature of C, we can then deduce from 
the galvanometer deflection the temperature of S from this graph. 



u Oivmie*-*? on Scale 

Fig. 133 

Secondly, we require the rate of rise of the temperature of the 
disc when the radiator is put on. 

As soon as the temperature in the enclosure, B, has become 
steady, the box is placed over S, keeping the latter at the centre, 
and the galvanometer is read at equal intervals of time ; these 
may conveniently be every 5 or 10 sees. 

The table shows a record of observations made every 10 sees. 





















These results are plotted on a graph. 

Scale Divisions 

Fig. 134 


Draw a tangent to the curve and measure the value of -=- as 

close to A as possible, since errors soon arise by conduction from 
the silver disc. 


m and s are found in the usual way. It will be convenient 
to regard these as constants of the apparatus, and to record 
them on the occasion of making the apparatus. It is inconvenient 
to make determinations at each experiment. 

Or s may be measured by Regnault's Method from a piece 
of silver identical, with that in the apparatus. 

The Rise of Boiling Point of Solutions 

The object of this experiment is to determine how the tem- 
perature at which a solution boils depends on the concentration, 
to determine the rise of the boiling point above that of the 
solvent, and to compare experimental results with those given 
by the thermo-dynamic formula. 

If P denote the osmotic pressure of a solution, i.e. the pressure 
due to dissolving one gramme molecule of the substance in one 
litre of the solvent, S, the molecular weight of a salt, and s, the 
number of grammes in ice. of the liquid, we have : 

P _ 2R0 

s S 

where R is the value of the gas constant per gramme molecule, 
and is the temperature of the solution, i.e. for any particular 
salt and solvent and for a temperature 0, 

P ocs. 

All this is true for dilute solutions only. 

A calculation based on the second law of thermo-dynamics 
shows that the rise of temperature of the boiling point, T, is 
given by the formula : 

T- £?' 

where P is the osmotic pressure, L = latent heat of vaporization 
of the solvent, p = its density, and = temperature of boiling of 
the solvent. 

For purposes of the calculation it may be assumed for the 

case of water, p = — at boiling point. 

P = 22-3 x 10 6 dynes per gramme molecule per litre at o°C. 
If = B.P., the approximate value for P at temperature is : 

— - x 22*3 xio 1 dynes. 
273 ° y 


e is in absolute units, and the value for any particular 
concentration may be deduced by simple proportion. 

L, for water at ordinary pressures, may be taken as 537 X 4*2 
X 10 7 ergs, in which units it must be expressed for the equation. 

Suppose w grammes of solute are dissolved in W grammes of 

Fig. 135 

. W W 

The volume of the solvent is — c.c. or 

P 1000 p 


If M denote the molecular weight of the solute, we have 

... . 1000 pw . 

litres, i.e. w — gramme mole- 
cules per litre. 

w , , • W 

— gramme molecules m 

M 6 1000 p 


Thus the depression according to the theory should be : 

1900 pw JP0 ioooze>P0 
MW ' L P ~ MWL ' 

Verify this expression by taking several values of w. 

A modern form of apparatus for carrying out this experiment 
will be described. It is represented diagrammatically in the 
figure (135). A is a glass tube which holds the solvent from 
which proceed two tubes, B and C, provided with corks. 

The thermometer passes down B, and the bulb is immersed 
m the solution or pure solvent in A. 

Weighed quantities of solute may be added through C. 

Through the cork, H, passes the metal tube of the condenser, F, 
which is cooled by passing in water by one of the tubes, T, and 
allowing it to flow out by the other. 

The small tube, G, carries garnets, which are placed at the 
bottom of A, and which together with a short wire through the 
base of A, assist the commencement of boiling and tend to prevent 
superheating. A stands on a cylinder of asbestos, D, which 
rests on a plate, E, of pipeclay, and the liquid is heated by two 
burners placed below E. 

The thermometer is set and its reading taken when the 
pure solvent is boiling, and the difference noted when 
the solute is added and the solution boils. 
i We have first to weigh the vessel, A, when detached 
from F, and not carrying the thermometer. Add a 
weighed amount of solute, w, to a convenient quantity 
of solvent. After solution, and after taking the boil- 
ing point of the solution, F is again removed and 
also the thermometer, and A is again weighed. 

Since w is known, we can obtain W, the weight 
of solvent. Care must be taken not to remove any of 
the liquid with removal of the thermometer. The 
thermometer bulb must be placed in the solution ; if 
it is above it attains the temperature of the steam 
coming off from the solution, and this is not at the 
temperature of the solution, but at the temperature 
of steam appropriate to the prevailing pressure. 

The Beckmann Thermometer # 

The magnitude of the change of temperature is not 
very great, and great care is necessary to measure 
B it accurately. The most convenient form of thermo- 
meter for this purpose is one of the Beckmann 
Fig, 136 type. 


A common form of this thermometer is made of Jena glass, 
having an apparent coefficient of expansion for mercury of 

magnitude—^—. It is illustrated in fig. 136. 

A special feature is the large bulb, B, containing a compara- 
tively large quantity of mercury from which a fine capillary 
tube extends, bent as shown at A, where the capillary is widened. 

The mercury in the bulb being of considerable volume gives 
a large additional volume on expansion, and this causes a large 
movement in the fine capillary tube when the change of 
temperature round the bulb is only slight. 

If, however, the quantity of mercury in the bulb were fixed 
the range of temperature for which the instrument could be 
used would be very small unless the capillary were inconveniently 

The thermometer is used to measure a smalt change in 
temperature above or below some particular temperature. 
Suppose that we require to measure a small change of temperature 
above T . 

It would be best to arrange so that the mercury extended 
from the bulb up to the zero of the scale when at the temperature, 
T . A slight increase in temperature would then drive the 
mercury along the capillary, which is marked in hundredths of 
degrees at intervals of sufficient width to enable an experimenter 
to estimate to thousandths. 

Suppose now that it is required to read temperatures slightly 
above another temperature, T lt which is greater than T . If 
we could withdraw sufficient mercury from the bulb it would 
be possible to arrange that the mercury extended up to the zero 
mark at the temperature, T v and slight increases would drive 
the thread along the fine tube as before. 

Of course, the expansion per degree rise of temperature in the 
second case is not strictly the same as in the first, since the 
initial volumes are not equal in the two cases. 

In order to see how this effects the observations, let us suppose 
that the graduations on the stem are correct at t ° C, i.e. with 
mercury filling the bulb and extending to the zero mark, when 
the surrounding temperature is t °, a rise of temperature of i° C. 
would cause an expansion to the first degree mark above the zero. 

Let the volume of mercury at t ° under these conditions be 
denoted by V , and let the coefficient of apparent expansion of 
the mercury in glass be a. 

Then the volume of the capillary per i° is V oc. 

Now suppose that mercury is drawn off until at temperature, 
/, the mercury fills the bulb and extends to zero. The volume 
drawn off is V a(rf — t ). 


The remaining mercury had a volume at t ° of 
V {i-a(*-* )f, 
and on further heating of one degree would expand by an amount 

V o^i -ce(* -t )}. 

Since the volume of the capillary per i° is V a this expansion 
will be registered as {i — <x.{t — t )\ of a degree. 

The degrees on the seale are too large for the temperature t. 

A correction could be made by dividing the scale readings by 
{i — cc(t — t )\, and it would Abe sufficient to add to the 
observations an amount <x{t — 1 ) per degree registered on the 
scale to obtain the corrected rise. 

As has been stated a has the value - — , so that if we suppose 

the thermometer correct at 90 C. and we record a rise just 
above ioo° C. we should add to each degree an amount 

? = -0016 . 


This is an appreciable amount since we can record up to 
thousandths of a degree. 

The correction can be neglected when we are recording to 
hundredths of a degree. 

The widening of the capillary at A enables the change in the 
quantity of mercury in the bulb to be made. 

Suppose it is necessary to read a temperature between ioo° 
and 101 C. The thermometer is placed into a bath at a tempera- 
ture of about 102 C. to 103 C, and so that the mercury extends 
from the bulb up to the widened part when at this temperature. 

If a quick jerk is made the thread will break at the point, 
where the capillary enters the wider part. 

When cooled to about 100 C, i.e. when held in steam, the 
temperature of which may be determined by observing the 
atmospheric pressure, and referring to a book of tables, the top 
of the thread should sink to the lower end of the capillary. It 
will not lie exactly at zero, and the position occupied by the end 
at this known temperature is recorded. It may be necessary to 
vary the upper temperature when the thread is continuous before 
the end will sink conveniently towards the zero at the temperature 
of steam. 

When this condition has been attained the small rise of 
temperature is readily recorded. 

It is not necessary to bring the mercury thread exactly to zero 
before reading the small temperature change, and if the correction 
is to be applied it may be made by giving the value to t which 
corresponds to the temperature for which the apparatus is set, 
just as if it had been set at the zero. 


If it is more convenient to observe a diminution of temperature 
below a certain point, it is only necessary to bring the top of 
the mercury thread at this temperature to a point high up on 
the scale. A slight depression will cause the top of the column 
to sink, and the interval may be measured as before. 

In such accurate temperature measurements it is necessary to 
make a correction for the emergent column. This is done by 
reference to Grutzmacher's tables taken from the " Zeitschrift 
fur Instrumentenkunde " (1896), p. 220. 

A reference to this table will show how important such a 
correction is. 






















—35 to —30 



to 5 




45 to 50 




95 to IOO 




145 to 150 




195 to 200 




245 to 250 




The second column gives the correction due to the variation 
of mercury within the bulb, and it shows that the thermometer 
reads correctly at a temperature between 5 and 45 , for in this 
region lies the point where the value of one scale division is i° C. 

The fourth column gives, in addition, the correction for the 
fact that part of the mercury is at a different temperature from 
that in the bulb. v 

In order to make a correction, choose the part of the table 
which describes most nearly the condition of the experiment, 
e.g., at a temperature in the neighbourhood of ioo° with a 
mean thread temperature about 30 , one degree recorded by 
the apparatus is approximately 1-032°. 

Two thermometers will be required, one adapted for use in 
the neighbourhood of o°C. and the other adapted for use near 
ioo° C. 
Molecular Weight by Depression of Freezing Point 

The thermo-dynamic formula for the depression of freezing 
point is 



where L is the latent heat of fusion of the solvent, p its density, 
and is the temperature of fusion. 

P is the osmotic pressure. 

Let W denote the weight of solvent used, and let w grammes 
of solute be added of molecular weight, M. 

Let P denote the pressure for one gramme molecule per litre, 
i.e. 22-3 xio 6 dynes/sq. cm. By using this value and p = i, 
L = 80 x 4*2 X 10' ergs, 6 = 273 Absolute, calculate the 
corresponding depression T . 


In the present case the volume of solvent is litres, and the 

r 1000 p 

number of gramme molecules dissolved is^rv m 

The corresponding depression is T 1 , where 

looop w Pe __w_ . I000 .Pf . 
1 " W M L P ~ WM lou L 

It will be the object of the experiment to calculate M. 

The vessel, D, with the stirrer, S, is weighed and a quantity 
of the solvent introduced, and the vessel again weighed. 

This gives W. 

The thermometer is placed with its bulb in melting ice, and 
the top of the mercury thread adjusted so that it stands near 
the upper end of the scale. This temperature is o° C. and small 
differences from this point may be read from the thermometer 
graduations. These may be corrected by Griitzmacher's Table. 

A weighed quantity of the solute is introduced into the solvent 
through the side tube, C, and solution brought about by help 
of the stirrer. 

The vessel is placed in an enclosure, E, and surrounded by a 
freezing mixture, F. 

The thermometer, B, must be kept in ice until D has cooled 
down to o°, and then the transference must be quickly made 
to D. Otherwise the thread will rise beyond the scale, and 
mercury will flow into A. 

During solidification the mercury will stand at a definite mark 
if the solvent is pure, and the difference, T 1 , can be measured. 

In order to set the thermometer ready for use in the experiment 
it must be placed in water cooled to within a degree or two of 
the freezing point. The thread of mercury which must be 
continuous from B to A is broken by a quick jerk at the top of 
A, when the temperature of the water is attained. 

In this way the thread will extend nearly to the top of the 
scale, when the bulb is at the temperature of melting ice. 


For the temperature at which the solution freezes the fall 
below the freezing point of water is obtained by subtracting the 
second reading with the bulb in the solution from that with the 
bulb in ice. The degrees thus read will require correction on 
account of the emergent column. 


Fig. 137 

In order to make the correction, use Grutzmacher's Table, 
e.g., if a temperature difference recorded is 1-5° C. with an 
average room temperature 15 C, the degree as read on the scale 
is really only -995°. Thus the difference is : 

i-5 x -995 = 1-492°. 

I ° T ° 

The thermometer reads to , so that we can estimate— — 

100 1000 

and it is therefore necessary to take account of such an error 
as that arising above, viz. of -008° C. 

Determination ot the Mechanical Equivalent of Heat by Friction 

The apparatus consists of two vessels of gun-metal of the 
shape of truncated cones, one fitting within another. The 
inner cone contains water, and is helcj in position by a measured 


couple applied by means of a weight hanging over a pulley and 
connected to a disc attached to the cone. 

The outer cone is rotated by means of a cord to a large wheel 
rotated by hand. 

Fig. 138 

It is possible by steady turning of the handle to cause the 
weight to hang almost steady, so that a steady couple is applied 
to the inner cone by friction, and its amount is measured by 
multiplying the force and the diameter of the wheel to which 
the cord is attached. 

A counting device is fixed to the outer cone so that the number 
of rotations- is recorded. 

If n is the number of rotations, w the weight in dynes applied 
to the disc, and a its arm, the work done is 

2tfnwa ergs. 

The rise in temperature of the water is measured by means 
of a thermometer, so that if M represents the total water equiva- 
lent of the cone and contained water, and if the rise in temperature 
is T, the heat developed is MT calories. 

If no heat is lost by radiation and conduction, the mechanical 
equivalent, J, or the number of ergs necessary to produce one 
calorie is given by : 


J = 


A common form of apparatus in use in laboratories is shown 
in the figure, it was designed by Dr. G. F. C. Searle. 

An improved design is described in the " Phil. Mag.," Sept. 
1920, by H. P. Waran. 

In order to prevent radiation losses it is a good plan to cause 
a rise in temperature of not more than io° C, and to cool the 
water about 5 below the room temperature initially. The two 


cones are mounted in a metal case lined with cork to diminish 
conduction losses. 

The heat is conducted through the walls of the inner cone to, 
the water, and it is necessary to stir the water during the experi- 
ment to assist the flow of heat throughout its extent. This 
may be done by means of the thermometer. 

One of the features of Waran's improvement is the automatic 
stirring of the liquid which consists of an oil with good 
conductivity and known specific heat. 

The Determination of Joule's Equivalent of Heat by Callendar and 
Barnes's Electrical Method 

The principle of the experiment is to supply electrical energy 
to a wire surrounded by water, and to measure the heat developed 
in the water by noting the rise of temperature. 

Fig. 139 

If we express the electrical energy in ergs or joules We can 
thus deduce that required to generate one calorie. 

Inside the glass tube, H (fig. 139), is fixed a helix of manganin 
wire of resistance about 9 ohms. The ends of the wire are 
joined to the terminals, C and C 1 . 

Water from the tank, B, enters H by a side inlet tube at 
one end and after flowing round the wire comes out from a 
similar outlet tube at the other. Thermometers, which enter 
the ends of H as shown, penetrate the inflowing and outflowing 
water, and serve to record the temperatures of the water before 
and after it has received heat from the wire. 

The vertical tubes, AA, allow air bubbles which may flow in 
with the water, to escape. The level of B is variable for the 
tank is movable up and down its stand. 

The makers of the apparatus recommend that the rate of flow 


should lie between 55 c.c. and 65 c.c. per min., or about 1 c.c. 
per sec. B should be adjusted to produce this rate. 

The tank consists of an inner and outer chamber. 

Through the tube, E, water enters the outer chamber from the 
supply, and comes out through M to enter H. 

The water flows over from the outer chamber to the inner 
and escapes to the drain by means of F. Thus water is supplied 
at a constant pressure to H, and the flow is made steady. The 
outflowing water may be collected in a measuring glass, 
or, of course, it may be collected in a weighed beaker and the 
mass per second flowing out deduced. 

The apparatus gives best results for a current of approximately 
2 amperes, so that a voltage of from 20 to 25 should be applied. 
To obtain a steady source ten or twelve accumulators should be 
used and connected through an adjustable resistance to C andC 1 . 

In this case, with the rate of flow indicated, a difference of 
temperature of the ingoing and outgoing water of about 8° C. 
is maintained. 

If possible the mean temperature of the water should be that 
of the room, to avoid errors due to radiation. 

When it is not possible to arrange this exactly, it is necessary 
to apply a correction. Let the temperature of the room round 
the apparatus be t , and the mean temperature of the water at 
entrance and exit be t. 

Then the number of calories lost per second in radiation is 

m{t — Q x -05, 

where m is the outflow per second. Or we may make the 
correction by adding -05(t — t ) to the difference of temperature 
recorded by T and T 1 . This correction has been obtained 
experimentally by the designer. 

Measure the resistance of the spiral by means of a Post Office 

Connect the resistance, R (fig. 140), in series with the supply 
of current and the ammeter, A, which has a range up to 3 amperes. 

If a voltmeter is available there is no need to measure the 
resistance of the helix, for the difference of potential between 
C and C 1 may be measured by connecting it through the key, K, 
in parallel as in the diagram. 

Adjust the current to about 2 amperes, and if necessary make 
slight variations from this to adjust the temperatures so that 
radiation losses are made as small as possible. 

Let m denote the rate of outflow per second, C the current in 
amperes, E the voltage supplied between C and C 1 , and T the 
difference in temperature of the two thermometers, corrected if 
necessary according to the rule given above. 


The supply of electrical energy is EC joules per second, and the 
heat developed is ml calories per second. 
Thus, J, the heat equivalent, is given by: 

JwT = EC, 

or J =^ joules" 

EC ».tuuuuuuub>- 

++■-+- GLh 

Fig. 140 

If the resistance of the wire is measured, instead of EC we 
must write C 2 R. The accuracy to be expected is between 
one-half and one per cent. 

The following is an example of an experiment carried out in 
the laboratory, and indicates the order of the quantities : 

Temperature of room, 17-3° C. 

Temperature of water at inlet, i6*8° C. 

Temperature of water at exit, 25-35° C. 

Mean temperature of water, 2i'07° C. 

Temperature difference of water hi the two cases, 8-55° C. 

Radiation correction = -05 x (21*07 ~~ I 7*3) = ' x 9° C. 

Corrected temperature difference, 8-74° C. 

Volume of water flowing out in 2 mins., 115 c.c. 

m = — - gms. per sec. 
120 ° r 

Current, 2 amperes. Resistance of wire, 8-92 ohms. 

,\ J = -J 2— = 4-260 joules. 

— x 874 
120 '^ 



The Sextant 

The instrument consists of a graduated arc, SS (fig. 141), with 
two radial arms, A and C. 

A third arm, B, moves about an axis through one of its ends 
at right angles to the plane of SS. It is fitted with a clamp 
and tangent screw, so that it can be accurately adjusted, and 
carries a vernier at its end which moves over the scale of SS. 

Fig. 141 

A plane mirror, M lf is attached to B, lying with its surface 
in the direction of B, and in a plane normal to that of the scale. 
The axis about which N B turns lies in the surface of this mirror, 
which is called the index glass. 

The second mirror, M 2 , fixed to the arm, C, is called the horizon 
glass, and its plane must also be perpendicular to the scale. 
It consists of a plate fif glass only one-half of which is silvered. 

At T, on the arm, A, is fixed a telescope with its axis parallel 
to the plane of SS and passing through the centre of M 2 . 

Suppose the movable arm is turned so that the mirrors are 
parallel, and T directed towards a distant object, on which it 
is focussed. 



Only one image will be seen of the object, for the light from 
M x and M 2 is brought into the telescope in the same direction, 
M 2 T. The rays by the two reflections may not he along the 
same line as those seen directly through M 2 , but since they are 
parallel there is only one image formed by the telescope (see 
fig. 142). 


Fig. 142 

If the index glass be rotated through an angle, A, by turning 
the arm, B, then the rays reflected by it into the telescope no 
longer come from the* same object as that which supplies the 
ray, QM 2 . Two superposed images are seen in the telescope, 
and the angle between PM^ and QM 2 is 2A, since on turning a 
reflector through any angle a beam of light is rotated through 
double this angle. Thus every degree on the scale, SS, corres- 
ponds to a difference of direction of two degrees. The scale is 
marked to give directly the angle between the rays, PMj and 
QM t , i.e. to measure the angle subtended at the instrument by 
two distant objects. 

In order to measure this angle one of the objects is observed 
directly through M 2 , while the other is made to produce a super- 
posed image on that of the first by rotating M x into the proper 
position. The angle through which the index glass is turned 
from the parallel position is then one-half of the angle subtended. 
Before making any measurement the instrument must be tested 
to see that it satisfies the following conditions : 

(1) The plane of the index glass must be normal to the plane 

of the scale, 

(2) The axis of the telescope must be parallel to the plane of 
the scale, and 

(3) The index and horizon glasses should be parallel, and at 
the same time the vernier should read zero. 

It will not be necessary, as a rule, to adjust for the first of 
these, but in order to see that the instrument is satisfactory in 
this respect look at the image of the scale in M^ Since M x 
passes through the centre of this scale the latter and its image 
will appear to intersect at the edge of the mirror and, if the- 
adjustment is satisfactory, to lie in the same plane. 

Both M x and M a are attached to frames which can be turned 


through small angles by means of screws. If necessary Mj may 
be adjusted until the test is satisfied. 

The second condition is tested by observing two objects and 
causing them to coincide at the centre of the field of view. The 
axis of the telescope is a line joining the centre of the object 
glass to the centre of the eyepiece, or to the centre of the field 
of view. Perpendicular to this axis lies one of the cross-wires. 
Tilt the instrument until the images lie near the edge of the 
field, and note if they still coincide. Then tilt it so that they 
lie near the opposite edge. If coincidence persists the axis is 
correctly adjusted. If this is not the case the telescope can be 
adjusted by means of the screws. 

Observe an object through the telescope and make its image 
appear in the field of view by reflection in M x . If it is possible 
to cause them to coincide, the two mirrors are parallel, and on 
account of the first adjustment this will mean that the third 
condition is partly satisfied. By means of the screw attached 
to M 2 the two images can be made to coincide if they do not do 
so at first. 

When these conditions are satisfactorily arranged it will 
probably happen that the pointer does not read zero when a 
distant object is viewed. To correct for this it is only necessary 
to note the zero error and apply it in all the observations. 
Coloured glasses are provided for diminishing the brightness of 
any object such as the sun. These can be made to intercept 
the light immediately before falling on the mirrors. 

Experiment i 

Place two candles at as great a distance as is convenient, and 
measure the angle they subtend at the instrument. 

Also find the angle by measuring the distance to each candle 
and their distance apart. 

Let the candles be C a and C 2 , and let S denote the sextant. 
The distances to be measured are a, b, and c, and if 
s = \{a + b + c) 


tan J-J(* -»>('-*)■ 
2 M S ( S _ a ) 

Check the values of obtained by the two means. 

Experiment 2 \ 

Let a trough of mercury or a carefully levelled mirror be 
placed so that the image of a lamp can be seen directly, and 
also by reflection, and measure the angle subtended at the 
sextant by the object and its image by holding the plane of the 
instrument vertically, at as large a distance from the lamp as 

The elevation of the lamp is half this angle. 


Fig. 144 

The diagram (fig. 144), illustrates that the angle, CBD, is 
measured since the instrument is of necessity above the surface, 

Actually, we require the angle, LAL 1 , but since we use a 
distant lamp, the two angles do not appreciably differ. 

Thus, the elevation may be measured by half the angle CBD. 

Measure the horizontal distance between L and A, and deduce 
the height of L above the floor. 

Check the result by actual measurement. 

Measurement of the Angles of Crystals by Wollaston's Goniometer 

The goniometer is a convenient instrument for measuring 
accurately the angles between the faces of small crystals which 
are too small to be examined by means of a spectrometer. 

It consists of a circular circle, S, which may be rotated by 
the large milled head, B, and its position read off by means 
of a fixed vernier, V. 

The crystal is fixed by soft wax to a plate, P, carried by an 
adjustable support, D, which may be rotated by the smaller 
head, A. 


The edge of the crystal formed by the two faces between which 
it is desired to measure the angle is adjusted so that it lies parallel 
to the axis of the circle. 

This adjustment is first made approximately by eye. In 
order to make the adjustment accurately, view the upper corner 
of a distant window in both faces. On rotation the images will 
move in a vertical plane if the edge is parallel to the axis. This 
may be tested by noting if each image moves in a direction 
parallel to the edge of the window as seen directly by the eye. 

Place the instrument so that the axis is parallel to a tall, 
distant window, and turn the screw-head, B, until the graduated 
circle comes against the stop. 

Fig. 145 

The eye is placed close to the crystal so that an image can be 
seen by reflection in one of the two faces. 

The axis is then rotated by the smaller screw-head, A, until 
the top of the window, as seen by reflection, appears to coincide 
with the bottom, as seen directly. 

When the adjustment has been made the angular position of 
the circle is noted. 

By means of B, the circle is now turned until the top of the 
window, as seen by reflection in the other face, coincides with 
the bottom, as seen directly. The second face now occupies 
a position parallel to the first, and if is the angle between 
them, the circle has been rotated through its supplement. This 
angle (180 — 0) is read off from the circle and deduced. 

It should be noted that the crystal must lie as close to the 
axis of the goniometer as possible, for the motion of the crystal 
from the first position with reflection in one face to the second 
position, with reflection in the adjacent face, consists both of 
translation and rotation, unless the crystal is on the axis. The 
amount of translation may be sufficient to cause an error in the 
angular measurement. 

If the window is a long way away the error is only small. 


The Determination of the Radii of Curvature of Spherical Mirrors 

(A) Concave Mirrors 

The most convenient method of determining the radius of 
curvature of a concave mirror is to place a pin point in front of 
it and to locate the position in which the image of the pin appears 
to coincide with the pin itself. The method of parallax is 
employed to ascertain when coincidence is attained. 

" The rays from the point of the pin falling on the mirror are 
reflected back from the surface along their original paths and 
must therefore strike it normally ; consequently, the pin point 
lies at the centre of curvature of the surface. 

Another method consists in locating a series of pairs of con- 
jugate points for the surface and using the formula : 

z -+i = * (I) 

v u r v ' 

A pin is set up as object and another pin adjusted until the 

image of the first coincides with it. We can then measure a 

pair of values, u and v. 
Several pairs of values are obtained, and the above formula 

then gives r. Take the average of four or six observations. 

(B) Convex Mirrors 
Method 1 

In the case of a convex mirror the image is virtual, and it is 
not convenient to locate it by a pin placed in a particular position 
since the image lies behind the mirror. 



Fig 146 

A pin, P, is set up in front of the mirror, CC, and in between 
them is placed a plane mirror, M, so that the image of P in both 
can be observed. The mirror is adjusted until the two images 
coincide (fig. 146.) 

By the simple law of reflection in the plane mirror, M, we 
know that the image of P in M, say, Q, lies at the same distance 
from M as P does, but on the other side of it. 

We can thus calculate the distance, AQ, for 
AQ = MQ - MA = MP - MA. 


Then v = — AQ, adopting the usual sign convention, viz. 
directions measured from A towards the object are positive, 
and in the opposite direction they are negative. 

Thus, using the formula (i) we can again deduce r by measuring 
AP and AM. 

Several pairs of values should be obtained, and they should 
give the same value of r. 

Method 2 

Another method is to set up a pin and form a real image of 
it by a convex lens. The image is located by placing a second 
pin, Q, so that there is no parallax between it and the image. 
Then place the convex surface between the lens and second pin, 
and move the surface until an image is formed coincident with 
the first pin, P. The rays after passing through the lens are 
directed to the point, Q, but strike the surface normally, and are 
therefore reflected back along their path. The radius of curvature 
of the convex surface is MQ (see fig. 147). 

Fig 147 
Method 3. By means of a telescope, metre rule, and small 

millimetre scale 
The diagrams (figs. 148, 149, 150) show the arrangement of 
apparatus. S is a small scale placed horizontally in contact with 
the surface of a convex mirror, along a line dividing it into two 
equal parts. 





Fig. 148 

The scale, RR 1 , is mounted at some convenient distance away, 
usually about 60 or 70 cms., and below its middle point is fixed a 
telescope, T, focussed on the image of the scale, RR 1 , in the mirror. 

The apparent length of the image is read off by means of SS 1 , 
which will be sufficiently well focussed to make this possible. 

The distance from the centre of the mirror, P, to the middle 



of RR 1 , is measured, say, d, and from these two measurements, 
together with the length of RR 1 , 2I, it is possible to calculate 
the radius of curvature of the surface of the mirror. 

Let rays from R and R 1 strike the mirror at L and L 1 , and be 


± T 


Fig. 149 

reflected down the telescope at O. Then LL 1 will denote the 
extent of the image, and the point, B, at which these two lines 
meet OP will be conjugate to O for reflection in the mirror. 
We may say that a point source at O will have a point image 
at B, so that if PB = x, we have : 

— J -uJ — _ 2 
% a r 

r denoting the numerical value of the radius of the mirror. 

R Fig,. 150 

But if we take LL 1 as approximately straight, since the image 
is of small dimensions, and denoting LL 1 by 2c 
I _ d + x 

C X 


so that 

x d\c J' 


2 _ 

r = 


I — 2C 

The result may be verified by means of a spherometer. 


The Focal Lines formed by a Coneave Mirror 

When light diverges from a point and falls on a mirror, it is 
supposed in the elementary theory that after reflection the rays 
pass through a single point or appear to proceed from a point. 
This is approximately true if the dimensions of the mirror are 
small compared with the distance from the source. A closer 
approximation to the truth is that the rays after reflection pass 
through two lines or appear to come from two lines, situated 
in parallel planes and perpendicular to one another. 

These are the focal lines and it will be shown how to calculate 
their positions theoretically, while it will be the object of experi- 
ment to verify the result obtained. 

We shall take the case of light falling obliquely on a concave 

If the point source lies on the axis of the mirror we have 
symmetry about this axis, and the two lines degenerate into a 
point or circle through which the rays pass. 

Fig. 151 

In the diagram, MM 1 denotes the concave mirror, and C its 
centre of curvature. The complete circle of which the section, 
MM 1 , forms a part is drawn for convenience. 

P denotes the position of the point source of light, and the 
diameter is drawn through P. 

The extreme rays, PM and PM 1 , are drawn and the reflected 
rays, MB, M X A, are drawn intersecting at F 1? and cutting off 
from the diameter the strip, AB. 


The mirror is a part of a sphere, so that rays falling on the 
mirror from P, whether in the plane of the figure or not will 
pass through AB. AB is thus a focal line; and is denoted by F 2 . 

If we imagine the figure to be rotated about the diameter, the 
rays, M X A and MB, will still intersect at a point but now out 
of the plane of the diagram. For a small rotation the point of 
intersection would be on a line through F x normal to the figure. 

Thus, a second focal line is through F x perpendicular to the 
plane of the figure. 

Let the angle of incidence at the point M 1 of the mirror be i, 
and let PD = p. 

Denote the distances of the focal lines from D by p ± and p 2 , 
i.e. DF X = Pl , and DF 2 = p 2 . 

Let the radius of curvature, CD, be R. 

Let the mirror subtend an angle, r, at C, a at P, and p at F v 
We shall regard its dimensions as small in comparison with R, 
P and pi, so that these three angles will be small and may be 
measured by drawing perpendiculars from M on to the corres- 
ponding lines, and dividing this perpendicular by the distance 
from the point concerned, 

MM 1 cos i 
e.g. a = , 


since the normal from M, on the line, PM 1 , makes an angle 
very nearly equal to i with MM 1 , and MM 1 is small and is 
regarded as straight. 
In the same way 

_ MM 1 cos i 

P ~~ "* 

(for cos i = cos (i + di) when 6i is small), 

,., MM 1 

while r = — — • 


From the triangles OMFi and POM 1 

^OPM 1 + ^OM x P = ziOFiM + z.OMF x 

.'. a + 2* = 2(» + di) + p, 

or. 2<5*' = a — 0; 

and in the same way from triangles, CQM and PQM 1 , 

M ' = a — r, 
;. 2f = a + 0, 

2MM 1 

or — — - = cos 1. MM 



\P Pi/ 

(- 1 + ~) 

\P Pi/ 

i.e. I - -\ 1 cos * = ^ 


This is the formula concerning the position of the first focal line. 
In order to determine p 2 , we note that : 

A PM X A = A PM J C + A CM^A, 
i.e. | • pp 2 sin 2,i = \ pR sin i + | Rp 2 sin *', 

which may be rewritten as : 

2 cos i i i ^ 

R p p 2 

So long as i remains constant, p x and p 2 vary with p just as 
u and w vary together in the formula : 

- + -=-• 
v u f 

When * is made zero we obtain the usual formula, 

- +— = * , 

p Pi r' 

for both Pi and p 2 , so that the two lines coincide. 

To find p x and p 2 experimentally, use a small hole in a screen, 
with a lamp behind as a source of light. 

Find R first by adjusting the mirror so that an image is formed 
on the screen at the side of the hole as in the experiment on 
concave mirrors, page 263. The distance from screen to mirror 
will then be R. 

Allow the light to fall on the mirror at angles of about 20 , 
30 , and 40 , and find the positions of the focal lines by means 
of a sheet or card or white paper held in a clamp. 

Measure the distances from the mirror to these lines, thus 
obtaining p x and p 2 . 

In order to find i, it is convenient to mount M on a stand 
carrying a pointer moving over a scale of degrees. 

Read off the mark against the pointer when the image of the 
hole is thrown back near the object, and turn from this position 
to any required incidence. 

If no scale of this kind is provided, measure the distances, 
PF X and PF 2 , say, a and b. 

We then know three sides of each of the triangles, PMFi and 
PMB, and can determine i from the usual trigonometrical formula 
for the tangent of half the angle of a triangle. 

Compare the values calculated in this way with those deduced 
from the formula derived from the above theoretical considera- 

Searle's Methods of Determining Optical Constants 

Accurate methods for the measurement of radii of curvature 
of polished surfaces, of focal lengths, and for the localization of 
the cardinal points, have been described by Dr. G. F. C. Searle. 



For the original accounts the reader is referred to the " Philoso- 
phical Magazine," Feb., 1911, pp. 218-224, or to the "Proceedings 
of the Optical Convention, 1912," pp. 161-172. 

The Determination of the Curvature of Spherical Surfaces 
For this purpose a table is mounted on a tripod stand (fig. 152), 
two of the feet of which carry screws for levelling. The table is 
horizontal, and can rotate truly about a vertical axis. It carries 
a millimetre scale on the top which can be clamped in any position, 
and a carriage bearing the surface slides along it. The figure 
shows a lens system in the place where the carriage slides. 

Fig. 152 

It is essential that, as the carriage slides along the scale, the 
centre of curvature of the surface should move along a line which 
intersects the axis of rotation of the table. 

In the figure the scale is shown, and also the wooden slider 
which acts as the carriage. On the slider is a metal mount 
which may be screwed to the carriage. This mount carries a 
horizontal spindle, each end of which is turned to a conical point. 
One of these ends is provided with a screw thread, so that it 
will fit a brass plate carrying three screws. 

The screws fit into a second brass plate to which the lens or 
mirror to be examined is fixed by a little wax. The arrangement 
thus provides a convenient means of adjusting the surface. 

The edge of the carrier is smooth and straight, so that it can 
slide along the scale, and an index mark serves to record the 

In preparing the apparatus for use the spindle is first set 
accurately parallel to the edge of the carrier. 

The tip of a pin is held just in contact with one of the conical 


points of the spindle, from which the brass plate is removed. 
The carrier is taken from the table top and replaced so that the 
other conical point lies near to the pin. 

If it is possible to bring this point and the pin into contact, 
the spindle is parallel to the edge of the board and scale. 

The spindle is rotated until this adjustment is possible, and 
the mount then firmly clamped to the carrier by means of the 
screw. It is also necessary that the axis of the spindle should 
intersect the axis about which the table turns. 

The scale is adjusted and the spindle moved along it until one 
point lies as nearly on the axis as can be judged by eye. 

A microscope is then brought up and focussed on the point, 
with the scale lying normally to the axis of the microscope, 
The table top is turned through 180 , so that if the axis of the 
table passes through A (fig. 153), and P denotes the first position 
of the spindle point, the second position will lie at P', where 
PM = M'P'. The slider has to be moved through the distance, 
P'P," in order to bring the point once more on to the cross-wires. 
If this distance is recorded and the slides moved back half this 
amount, the point will be at M', the foot of the perpendicular 
from the axis to the line P'P". 


M 1 
— 1 — 





.M u 

Fig. 153 

The microscope is then traversed so that it is focussed on the 
point. Its axis is thus directed along MA. Turn the table 
through 90 , so that M' moves to M" ; and move the scale along 
a direction perpendicular to itself until the point once more 
comes on the microscope cross-wire. The point then lies on the 
axis of rotation of the table, and the axis of the spindle will 
intersect that of the table as it slides along the scale. 

Mount the surface to be examined on the brass plate, and 
attach it to the spindle. If the surface is one of the faces of a 
lens, the side not under examination should be blacked by 
covering it with vaseline and lamp black to absorb rays striking it. 

Set up an object in some convenient position and view its 
image in the surface. If on rotating the spindle about its hori- 
zontal axis the image does not move, the centre of curvature 
lies on this axis, and the preliminary adjustments are complete. 
Since the image remains stationary the rotation of the surface 
serves to bring up fresh parts of the sphere exactly into the 
place occupied by the part moved away, i.e. the centre lies on 
the axis of rotation. 


The object may conveniently be a fine brightly illuminated 
line drawn on a piece of ground glass. 

Set up this object again so that rays fall nearly normally on 
the surface, and examine the image formed by reflection in it 
by means of a microscope. By moving the carriage a position 
can be found in which a rotation of the table to and fro produces 
no displacement of the image. This occurs when the centre of 
curvature of the surface lies on the axis of revolution of the 
table, for rotation then has the effect of replacing One part of 
the surface by another, and reflection occurs as before. 

The position of the index mark is noted. 

If the lens is moved until the vertical axis of the table is a 
tangent to the surface, a rotation to and fro will not displace 
the centre of the lens. Place in the centre a few grains of lyco- 
podium powder, and focus the microscope on one of these. 
The lens is moved and the focussing repeated until the movement 
ceases. The index reading is again noted. The difference of 
the two gives the radius of curvature. 

Preliminary observations may be made by eye until the motion 
appears to cease in the two cases, and only the final exact adjust- 
ments need be made by the aid of the microscope. 


Introductory Remarks 

Rays of light in passing from one medium to another usually 
undergo a deviation from their course in the first medium. 

On the wave theory this is accounted for by the fact that the 
light travels with different velocities in the two media. The 
refractive index with respect to two media is defined to be the 
velocity of light in the first divided by the velocity in the second, 
and we denote this by x yL 2 . 

When light travels from air to another medium, say glass or 
water, we shall write a \L g or n«, unless there is no doubt that we 
are dealing with air and some other stated medium when we may 
write simply \x. 

We have by definition, 


and i^s = ~' 


In particular 

■"" = ^T 

a result which will shortly be found useful. 

In many experiments in this and following chapters it will be 
necessary to furnish a bright source of monochromatic light. 
The most convenient way to provide such a source is to use a 
Mecca burner, which consists of a Bunsen burner rather larger 
than the ordinary type of burner provided with a wide end over 
which is stretched a gauze with a wide mesh. 

If a small bead of soda glass is placed on this gauze and the 
Bunsen made to roar as much as possible a quite satisfactory 
yellow flame will be produced. 

It is a great advantage that there is no crackling in the flame 
as in the case of the use of common salt, when small pieces of hot 
salt are thrown about falling on the bench and on the slits of 
spectrometers. In the case of the latter serious damage to their 
shape may result. 

The slit may be illuminated directly, or better still, an image 
of the brightest part of the flame may be thrown on to it by a 
short focus convex lens. 




The student will do well to pay attention to the small point of 
illumination of the slit. It is important always, but assumes 
greater importance in the case of experiments on interference 
and diffraction which will be described in the next chapter. 
The difficulty of discovering Newton's rings, interference fringes, 
and diffraction bands is almost always due to a lack of care to 
obtain the best possible illumination of the slit or whatever may 
be used as a source of light. 

Determination of the Refractive Index of a Plate by its Apparent 

The apparatus necessary is a good travelling microscope and a 
plane cover glass. 

Set the microscope with its axis vertical and focus it on the 
metal platform. 

Note the reading on the vertical scale. 

Insert the cover glass over the point on which the instrument 
was focussed, again focus on the metal and observe the scale 

Usually the metal surface, though dark, is easy to observe, but 
if desired a thin sheet of white paper may be placed over it and 
the surface of the paper used instead of the metal surface. 

Raise the microscope until the upper surface of the cover glass 
is sharply focussed. There will usually be specks of dust on the 
surface to assist this setting of the microscope, but if any difficulty 
arises place a small drop of ink on the surface and focus the 
extreme edge of the drop. The drop need be no larger than that 
made by a sharply pointed pen. 

Note the scale reading when this third adjustment is made. 
If we describe the scale readings by (1), (2), and (3) respectively ; 
the difference between (1) and (3) gives the actual thickness of 
the glass and that between (2) and (3) the apparent tfrickness. 

Consider a point, P (fig. 154), situated at the lower surface of 



the glass and from which rays of light originate. Only those 
making small angles with the normal to the surface will enter 
an eye of microscope placed above P. 

If one of these rays, PO^ 1 , makes an angle, *, with the normal 
in the air and an angle, r, in the glass, 

sin*" OO 1 0*P O x P 

H = 

so that \l = 7^7 = , ., . f 

sm r O 1 ? 1 OO 1 ~~ Q 1 ? 1 ' 

For small values of * and r we may write : 

O 1 ? 1 = OP 1 , O x P = OP, 
OP _ thickness of glass 
OP 1 — apparent thickness' 

for all rays inclined at such small angles appear to come from P 1 , 
so that OP 1 is the apparent thickness. 

Determination of the Refractive Index of Liquids by Total Reflection 

When a ray of light passes from a medium of refractive index, 
Pi, to another of refractive index, \t 2 , with an angle of incidence, *', 
and of refraction, r, we have : 

\x 2 sin * 
ll * 2 — nj ~~ sin r 
We may not always find a corresponding value of r for a given i 

unless — is less than unity. 

In the case when — > i, the value — sin % must not exceed 

{*2 t*2 

unity. In the limiting case when 

sm * = — , 

the corresponding value of r is 90 , and i then measures the 
critical angle. For values of i greater than this critical value 
the surface acts as a perfect reflector. 

If i is slowly increased a value is finally attained when the 
refracted ray suddenly disappears. 

In this case if the second medium is air so that \l 2 = 1 we have : 


— = sm *. 


This formula may be used to determine y. f by mounting a small 
rectangular trough, CD (fig. 155), with sides of plane glass on 
the table of a spectrometer, so that a parallel beam of light may 
be passed through it from a collimator, GH, and received in the 



telescope, AB. The light is suddenly cut off when the air cell, 
EF, consisting of two plates of glass mounted parallel to one 
another and cemented together with a thin air space between, 
is turned so that the light falls on the air at the critical angle. 



It will be noted that the critical angle is that for air and glass, 
but the apparatus is used to determine the refractive index of 
the liquid in the trough. 

Let the ray, ABCDE, be incident from the water on the glass 
and be totally reflected at the glass-air surface at the critical 
angle. Then if FBG and FTOG 1 be the normals to the glass 
at B and D, we have : 

— = sin z BCH, 

where CH is the normal at C and aV . g denotes the refractive 
index from air to glass. 



^aV-a _ sin ABF sin ABF 
av. w ~ sin CBG ~~ sin BCH' 

*'• olXw ~"sinABF '"( J ) 

We have thus to measure the angle ABF, and from it we deduce 
the value a [i Wf the refractive index from air to water. 

As a source of light use a Bunsen flame containing sodium* 
and illuminate the slit of the collimator, which must be adjusted 
for parallel light (p. 279). Focus the telescope on the slit and 
turn EF (fig. 155), until the light just appears. EF is attached 


to a pointer which moves over a scale of degrees. Note the 
position of the pointer. Turn EF from this position into another 
where the image again disappears. 

Let AA 1 denote the axis of the telescope and EF the first 
position. The second will be E X F X if L AOE = L AOE 1 . The 

Fig. 157 

second position is reached either by turning through the angle 
EOE* or EOF*, and ^EOF* = * - iLEOE*. 

When the apparatus is used there is no uncertainty concerning 
which angle is measured. 

The angle we require is that between the ray and normal to 
EF or ET 1 , i.e. the angle between OA and the normal to EF or 
E*F*. This angle is half the angle between the two normals, and 
this is the same as half the angle E x OF. 

Thus we have to note the angle through which EF turns and 
take one half of it to find the angle ABF of formula (1). 

To Find the Refractive Index of a Liquid, using a Lens and a Plane 

In the experiment a plano-concave lens is formed of the liquid 
under examination (fig. 158), and its focal length found experi- 
mentally. The refractive index and radii of curvature of the 
two surfaces enter into the formula for the focal length, so that 
it is possible to deduce the index, ja, from a determination of the 
focal length, /, and the radius of the curved surface of the lens. 
The liquid lens is made by placing a drop of the liquid on a plane 
mirror and laying a convex lens of from 10 to 20 cms. focal 
length on the drop. The liquid is squeezed into the space between 

the mirror and lens and we have a combination of two lenses 

one of glass and the other of liquid, giving a combined focal 
length of, say, F. 


If the convex lens has a focal length, /', then : 


We may therefore deduce / from a knowledge of F and/'. 
To determine /', place the lens horizontally on the mirror and 
adjust a pin, held in a stand above the lens, until the inverted 

Fig. 158 

image and object appear together and there is no parallax 
between them. The distance from lens to object gives /' (com- 
pare p. 263). Now place the liquid and lens on the mirror and 
again find the position of coincidence of object and image. 

This gives F, so that we now have /. 

For the purpose of substituting in the above formula, F and/' 
must be given their appropriate negative signs, and / will turn 
out to be positive. 

If r is the radius of curvative of the curved liquid surface, 
and n the refractive index of the medium, 

~f-~T~ < a > 

since the second surface has zero curvature. 

r can be found by measuring the radius of the surface of the 
lens in contact with the liquid by means of a spherometer, or 
by any of the methods described in the last chapter, so that n 
can be calculated from the equation (2). 

If it is preferred, the determination of r may be avoided. 

If water be used and its index regarded as known and having 
the value, 1-33, we may make a water lens as above, and calculate 
its focal length, /*. 

But 7=T-> 

or ix = 1 4- 


/ ' 
where p is the refractive index of a liquid other than water. 


A convenient liquid to use for the experiment is aniline. 
Care must be taken to prevent it from getting at the back of the 
mirror since it dissolves the varnish protecting the silvering. 

Determination of the Refractive Index of a Lens by Boys' Method 

This method of finding the refractive index of a lens consists 
in measuring its focal length and the radii of curvature of both 
surfaces. The value of {* may then be determined from the 
equation : 

T- <*-<}-!> 

where / denotes the focal length of the lens and r and s the radii 
of curvature of its surfaces, with the usual convention regarding 
the signs of the quantities in the formula. In a convex lens let 
v x and r 2 denote the radii of curvature numerically and F the 
numerical value of the focal length. Then : 

**-o-*(Z + k) 

Determine F by any of the methods described below, p. 293. 
In order to find r lt set up a pin with its point on a level with the 
centre of the lens. Two images will be seen by reflection in the 
faces of the lens, one erect from the front surface which acts as a 
convex mirror and one inverted by the concave back surface. 
It is the latter which is required for the experiment. Move the 
pin until its inverted image is coincident with it, as judged by 
the method of parallax. When this is the case the rays must 
strike the back surface normally and be returned along their 
incident course. 

If an eye be placed on the side of the lens remote from P 
(fig. 159), it will receive the transmitted part of the ray, PB, and 
will see the image of P in the direction, CBQ. 

C __ 


Fig. 159 

Q will thus be the image of P in the lens, and OP and OQ 
are conjugate distances. Thus, writing u = OP and v = OQ, we 
have : 

1 i_ _ 1 , . 

OQ OP F * [3) 


This enables us to determine OQ from OP and F. 

Q is the centre of curvature of the left-hand surface of the lens 
since QAB is normal to this surface. Thus OQ gives the value 
of the radius, say, r x . 

Turn the lens round and repeat the process to obtain r 2 , the 
radius of curvature of the other face of the lens. 

Let OP = d x in the first case, and let d 2 denote the corresponding 
distance in the second. Then from (3) : 


~1 x 



d 2 



= (*- 


U7 + 


d z 



so that we may calculate ^ from the experimental determination 
of F, d v and d z . 

It is sometimes difficult to see the image by reflection at the 
back surface of the lens, but by holding it over the surface of 
mercury or floating it in the mercury, the image may be made 
to stand out and be easily located. 

The Spectrometer 

The spectrometer consists essentially of a telescope and colli- 
mator. The latter is a system of lenses mounted in a telescopic 
tube with an adjustable slit at one end, and it serves the purpose 
of rendering rays from the illuminated slit parallel on emergence. 
Both are mounted on a rigid stand, the collimator being fixed, 
and the telescope rigidly attached to an arm which rotates about 
the centre of the stand. Both are mounted horizontally with 
their axes in the same plane. 



. I _ 

a .' t 

A 1 B 


Fig. 160 

The centre of the instrument is occupied by a table provided 
with screws for levelling. Underneath the table is a metal scale 
of degrees on which can be read off the positions of the telescope 
and of the table. In the diagram, T denotes the telescope, AB 
the table, and C is the collimator. PP denotes the axis of the 
instrument and the table and telescope rotate round it. The 
telescope is fitted with a Ramsden eyepiece carrying cross-wires. 


Before using the spectrometer for any experiment certain 
adjustments have to be made in addition to those made in the 
construction of the apparatus by the maker. 

In the first place, the eyepiece of the telescope is adjusted so 
that the cross-wires are distinctly focussed. The telescope 
should then be taken to an open window and focussed on a distant 
object such as a distant telegraph post, care being taken that 
there is no' parallax between the image and the cross-wires. 

When this is done the telescope must not be readjusted again 
during an experiment or it will be necessary to repeat this process. 
It may happen that a second observer whose sight differs from 
that of the first is unable to focus the cross-wires easily. He may 
readjust the eyepiece provided that he does not alter any other 
part of the telescope, for then he is not causing it to be out of 
focus for parallel rays. He merely gives himself convenience 
in focussing easily and leaves the cross-wires and image without 

The telescope is now turned towards the collimator, the slit 
being made vertical and illuminated with monochromatic light, 
and the collimator is adjusted until a distinct image of the slit 
falls on the cross-wires. 

The apparatus is now adjusted so that parallel rays pass from 
collimator to telescope. 



"ill - 



" -T- 

1 i -*—^ 

Fig. i6i 

When a prism is used in the spectrometer it is necessary to 
adjust it so that its refracting edge is vertical. 

The screws on the table enable this to be done. They are shown 
at D, E, and F, placed at the corners of an equilateral triangle, 
and the table is usually ruled with lines as shown (fig. 161), to 
assist in setting the prism with one face normal to one of the 
sides of the triangle, for example EF. 

The table should be made as nearly horizontal as possible by 
the use of a spirit-level, and the prism then placed with one face 
perpendicular to EF.' 


Now suppose light from the collimator falls on this face and is 
reflected into the telescope. If this face is vertical the slit will 
now appear to lie in the same part of the field of view of the 
telescope as when it is seen directly. 

The three screws should be adjusted to restore the image to 
its direct position if necessary. 

Fig. 162 
* Let the light be reflected into the telescope by the other face 
bounding the refracting edge. If further adjustment is necessary 
it must be done by the screw, D, for this will not disturb the 
previous adjustment, since it does not turn the face perpendicular 
to EF out of its vertical plane. 

The two faces are now vertical and the instrument is adjusted. 

It is sometimes necessary to arrange one face of a prism so 
that it lies normal to the collimator or telescope. 

This may be done by turning the telescope from the position 
in which the slit is seen directly without the prism, through a 
right angle, so that the axes of the telescope and collimator are 
perpendicular to each other. 

The prism is now placed on the table of the instrument and the 
table rotated until, by reflection in the face concerned, an image 
of the slit is thrown on the cross-wires. The face now lies at 45 
to the axes of the collimator and telescope and a further rotation 
of 45° will bring it either perpendicular to the collimator or 

In making measurements with the spectrometer the slit should, 
as a rule, be narrow, and the cross-wire should lie accurately 
down the centre of the image of the slit before the position of 
the telescope or table is noted on the metal scale. 


Tangent screws help in the accurate final setting of the telescope 
and table, and verniers on this scale serve to measure the angles 
to an accuracy of one minute of arc. 

A modern form of the apparatus is illustrated in fig. 162. 

The Refractive Index of a Prism by the Method of Minimum Deviation 

When a prism with a refracting angle, A, causes a minimum 
deviation, D, in light passing through it the refractive index is 
measured by the formula : 

sin £ (A + D) 

sin I A 

Thus the determination of jx consists in measuring A and D. 
There are two methods in common use for measuring an angle 
of a prism. 

One of these consists in allowing the light from a narrow slit to 
fall partly on one of the faces of the prism bounding the angle and 
partly on the other. An image of the slit can be seen in the 
telescope when the latter lies on either side of the prism. 

Thus in the case illustrated (fig. 163), the telescope will receive 
rays from the direction AO on one side and from AK on the other. 
The telescope has to be turned through the angle KAO in turning 
from one direction to the other. 

F D L 

Fig. 163 
Since DA and AO make equal angles with AC, we have : 
^- EAC = z. CAO, 
and similarly, /_ EAB = ^ BAK. 

Thus £_ OAK = 2 ^ BAC. 

To measure A we need only focus the image of the slit on the 
cross-wire in the two positions and halve the angle through which 
it is rotated. 


Suppose AB is the face which reflects the light into the telescope 
along AK. If the prism is rotated until the face, AC, now lies 
parallel to AB the rays will once more be reflected in the direction, 
AK, but now by AC. Some of these will enter the telescope if 
there is a sufficiently broad pencil of them. But the prism has 
been rotated so that AC moves round through the angle CAB 1 . 

Thus we measure, by means of the table, the angle (180 - A), 
and A can be deduced. 

Measure the angle of the prism by both methods. 

It now remains to find D. 

Set the prism so that with A as refracting angle, light is refracted 
through it and received in T. 

It will be found that as the prism is rotated the telescope has 
to be rotated to keep the image of the slit in the field of view. 
Rotate the prism until T is as close to the position directly 
opposite the collimator as possible, with the slit in the field of 
view. When this is the case the angle between the telescope and 
direct position is as small as possible. 

Note the position of T and then remove the prism and observe 
the slit directly. Again note T's position, and hence find D. 

It is, of course, necessary to adjust the spectrometer and edge 
of prism m the way described in the section on the adjustments 
of the spectrometer. 

The final movements of the telescope or table must be made 
carefully with the tangent screws. 

The Dispersive Power of a Prism 

The dispersive power of a medium is measured by : 

t*B — v.* 

w =- 

v- — 1 

t* B and n R are the refractive indices for blue and red rays and u 
has the value, £ (^ + ^). J * 

The refractive indices may be found by the method of minimum 

As a source of blue and red rays a discharge tube containing 
hydrogen may be used. 

The^ tube should be held vertically, and the slit illuminated 
by it directly, or an image of a bright part of the tube thrown on 
to it by means of a short focus lens. 


Three well-marked lines can readily be seen, one in the red, 
a second of blue-green colour, and a third in the violet. Use the 
first and third of these — they are known as the C and H r lines 
respectively, while the second is the F line. 

The Refractive Index of a Liquid by Total Internal Reflection within 
a Glass Prism 

A glass prism with one face unpolished is mounted on the table 
of a spectrometer. The table is levelled and the edge between 
polished surfaces set normal to its plane (see p. 281). The 
angle between these faces is measured in the usual way by allow- 
ing light from a wide collimator slit to fall on both faces and by 
measuring the angle between the two reflected beams. 

Light from a sodium flame is allowed to fall on the unpolished 
surface, or an image of the flame is thrown on it by means of a 
lens to cover the whole matt surface. 

It is first necessary to find the refractive index of the glass 
prism and then to coat one of the bright faces with a thin layer 
of the liquid. Glycerine is a very convenient substance with 
which to carry out the experiment. 

The theory of both parts of the experiment is the same. 


ABC denotes the prism, of which AB is the unpolished side. 
This side acts as a collection of point sources of light of which S 
denotes one. Rays from it strike AC and are reflected and re- 
fracted there. Those like SD falling at an angle of incidence 
less than the critical angle are partially reflected and partially 
refracted so that the ray, EF, issuing from BC is less intense than 
the incident ray, SD. 


Fig. 164 

Rays falling at angles greater than the critical angle suffer no 
refraction at AC, so that the emergent ray from BC is scarcely 
less bright than the incident ray. In any case there will be a 


marked difference between the former and latter group. The 
ray, SG, is drawn for critical incidence so that the direction of 
the emergent ray, HK, stands between those for the faint and 
bright rays. * 

We get a similar state of things for any other point in AB, and 
for each point the rays striking AC at critical incidence give rise 
to rays parallel to HK on emergence. 

Such a group of rays is brought to a focus in the focal plane of 
a telescope placed to receive them. 

Similarly, a group of rays, parallel to EF, will correspond 
to all rays falling on AC parallel to SD, and these will fall on the 
telescope in a direction different from that of HK, and will form 
a line in the focal plane not coincident with the former. This is 
true for all the directions of rays from BC. We shall thus have a 
multitude of parallel lines in the field of view divided into two 
groups of different intensity by the critical direction, HK. The 
effect will produce a field sharply divided into bright and dark 
halves by the direction, HK, making a with the normal to BC. 

If the telescope is turned to face AC the field will be similarly 
illuminated on account of the reflections that have taken place 
on BC. The issuing critical rays will make an angle, a, with the 
normal to AC. 

If the cross-wires of the telescope are set on the dividing line 
of the field when it is directed towards BC, and again when 
towards CA, we can deduce the value of a by observing the angle 
through which the telescope has been turned, since the telescope 
is rotated through 

180 — C + 2a = (say). 
Thus 6 + C ~ l8o ° 

a = 


where it is assumed that the telescope is moved from side, BC, 
towards CA in a counter-clockwise direction as seen in the 

From C and a we can calculate \l from the formula given below. 

Suppose that AC is coated with a substance of refractive index, 
Pi, and that the index of the prism is p. 

We have on referring to the diagram : 

v-i = v- sin c, 
p + c = C, where = z. MHG 
since the points C, G, M, H, are concyclic, 
sin a = {x sin 0, 
.*. l*i = V- sin (C — p) = {i sin C (i — sin 2 p)* — cos C sin a 
= sin C (n 2 — sin 2 a)* — cos C sin a. 


When (*! = i, i.e. when there is no layer on face, AC, we have : 

„ /r + sin a cos C V ■ • 2 
^2 = 1 -T- ) + sm 2 a> 

V sinC J 

This is the formula from which n may be calculated. 

Fig. 165 shows the state of affairs when the critical angle is 
large. This will be the case when the media on either side of 
AC are nearly of the same optical density. 


Fig. 165 

The formula above will still apply if the value of a has its sign 

The above method of measuring a can be used in finding {*, 
but when one side is coated it would be necessary to clean that 
side and coat the other when the telescope is turned. This would 
make it difficult to keep the prism fixed and would be 

It is therefore best to use an eyepiece, such as the Gauss eye- 
piece with cross-wires that can be illuminated. 

Fig. 166 

The light from these may then be reflected in the face opposite 
the telescope and the image made to coincide with the object. 
When this is the case the telescope stands perpendicularly to the 
face and the angle between this direction and that in which the 
division of the field is viewed measures a. 


After finding jx for the glass, place a few drops of the liquid on 
one of the polished faces and press over it a thin plate. This 
ensures that the face is covered with liquid, 

It is better for the sake of definition of the two halves of the 
field to allow light to fall at grazing incidence on the prism 
surface, say, AC. Then the rays entering the prism make angles 
less than c with the normal so that the field is now only half 
illuminated and the edge corresponds to the direction, HK. 
AB should be kept dark by covering with a sheet of dark paper. 

If the light is incident externally on the liquid film, it must enter 
by the edge, AD (fig. 166), any ray such as P would not reach AC 
at grazing incidence. 

Calibration of a Spectroscope 

The spectroscope already described had only one prism, but 
it is an advantage sometimes to use two mounted together on 
the spectrometer table. The experiment may be carried out 
with one only, but a wider spectrum is obtained with two. 

S .L 0. i %\ ^rj>k * 

Fig. 167 

If the prisms, P 2 and P 2 , are used to produce a series of spectral 
lines due to some source, S lt the telescope, T lf will have a definite 
position on its scale corresponding to each particular wave length. 
To calibrate a telescope is to find the values of the wave lengths 
corresponding to the different parts of the scale. If a curve is 
drawn whose co-ordinates are the wave lengths and corresponding 
scale readings, such a curve is a 'calibration curve,' and may be 
used to determine any wave length from the division of the 
telescope scale at which it is seen. 


First make the usual adjustments for parallel light. Place 
one prism, P x , in the position corresponding to minimum deviation 
and then put in P 2 also in the position of minimum deviation. 

Examine the spectrum of a Bunsen flame containing sodium 
light. Fix the telescope so that the sodium line lies on the cross- 
wire and note the position on the metal scale of the spectrometer. 

A graduated scale is provided consisting of close rulings cut on 
an opaque screen so that the lines are transparent. 

It is fitted at the end of a second collimator, C 2 , which is 
adjusted to direct rays down the telescope after reflection on 
one face of P 2 . 

The telescope is already set for parallel rays so that by 
adjusting C 2 so that an image of the scale, ss, appears at the 
cross-wires, it is ensured that parallel rays emerge from C 2 . 

The position of the sodium line is taken as a point of reference 
and scale divisions are read so many to the right or left. If the 
scale, ss, is moved accidentally no error is then caused and the 
calibration curve will still be of use. 

In a good spectroscope the sodium lines will be separated and 
will appear as two very close together. Use one of these as the 
reference line. 

Measure the scale positions of a series of lines of known wave 
length extending over the visible spectrum. 

Use only sharply denned lines. 

The table gives the wave lengths of lines which may be pro- 
duced conveniently in the laboratory. 

In order to produce the lines from the metals use a spark 
between poles made of these metals connecting them to opposite 
terminals of a Leyden jar charged by an induction coil. 

The salts are volatilized in a Bunsen flame as with sodium. 
Examine the inner cone of a Bunsen flame ; it is due to carbon 
monoxide, and contains several series of lines. The spectra of 
light from discharge tubes containing various gases should be 
examined as well. 

Lines from a Neon lamp should be examined also. The 
field will be observed to contain many lines, the yellow line of 
wave length 5853 units is a bright line and it is a good exercise 
to determine from the curve the wave lengths of the other lines, 
afterwards comparing with standard tables. 

Plot the curve on a large scale on squared paper. 

The Auto-Collimating Spectrometer 

In this instrument the telescope acts also as the collimator. 
The apparatus is almost identical with the ordinary spectrometer, 





Lithium Chloride 
or any Salt of 

Any Salt of 

Salt of Potas- 







Double Yellow 


Extreme Violet 


(Not to be confused with 
the bands) 




Red (C) 

Blue-Green (F) 
Violet (H r ) 









except that there is no separate collimator and the telescope is 

On looking into the eyepiece of the instrument, the field is 
seen to be divided into two parts, the lower half is dark and the 
upper bright, but crossed by a pin which extends from the upper 
edge downwards across about half of the bright part of the field. 

Near the eyepiece end the tube of the telescope is provided 
with an opening by means of which a slit just within may be 

Just below the dividing line of the two halves of the field of 
view a small right-angled prism is placed which deviates the light 
from the slit down the centre of the telescope tube. Its position 
is denoted by the dotted rectangle, P, in fig. 168, 


If a polished surface such as the face of a prism is placed in 
front of the object glass and set at right angles to the axis of the 
telescope we may, by focussing, make the rays leaving the object 
glass parallel, so that they will be returned by reflection at the 
face of the prism and come to a focus in the focal plane of the 



Focus the eyepiece carefully or^ the pin. Illuminate the slit 
by means of a small electric lamp — a four-volt lamp, supplied 
by two accumulators, is convenient for the purpose. The slit 
will be observed to be of the shape of an inverted T. The image 
must be brought so that the horizontal bar lies along the edge 
bounding the two halves of the field, when the end of the vertical 
bar will coincide with the point of the pin. In order to obtain 
this result the face of the exterior prism must be accurately 
normal to the emergent rays. The table is provided with three 
screws in order to level to the prism, and the process described 
on p. 280 must be followed. 

The positions of the slit and pin are such that when the image 
of the slit lies at the same distance from the object glass as the 
pin, the latter is at the principal focus. 

The prism is slightly tilted to throw the image of the slit on 
to the upper half of the field of view. 

To Find the Refractive Index of a Prism by means of the Auto-colli- 
mating Spectrometer 

Place the prism on the table and adjust the faces bounding the 
refracting angle as described above so that the light reflected 
normally by both throws an image of the slit into the field of 
view just below the pin. 

When this is so the faces are vertical and consequently so also 
is the edge of the prism. 


Use monochromatic light by placing a sheet of yellow glass 
between the bulb of the electric lamp and the slit. 

Observe the position of the table when the face, AB, reflects 
the light normally. Rotate the table until the light falling on 
AB and refracted there strikes AC normally and is returned along 

Fig. 169 

its path. The table has been turned so that the face, AB, has 
turned from a direction normal to the rays from the telescope 
into the position at which refraction takes place, i.e. it has turned 
through the angle of incidence, i. The angle of refraction in this 
case is the same as the angle, A, and is so marked in the diagram. 

This angle may be measured by setting the face, AB, normal 
to the rays and reading off the position of the table. Then by 
rotating the table until the rays strike AC normally we turn the 
table through an angle (180 — A) . 

We may therefore calculate ja, since we know the angle of 
incidence and refraction in a particular case and 

_ sin i _ sin i 
sin r sin A" 

The Pulfrich Refractometer 

This apparatus is shown in figure 170. It is designed to 
measure refractive indices of solids and liquids to an accuracy 
of about r~ per cent. 

The principal part of the apparatus consists of a prism having 
two plane polished faces at right angles to each other. One of 
these is horizontal and the other vertical. On the horizontal 
face is placed the substance whose refractive index is required. 
If this substance is a solid it must have two faces cut perpendicu- 
larly to one another, both of which are cut accurately plane, so 
that one may rest on the horizontal surface of the prism and the 
other stand vertically. Optical contact is brought about by 
placing a few drops of a liquid on the horizontal surface of the 


prism, which has a refractiye index higher than that of the solid 
to be experimented upon, and standing the solid on it. The 
makers recommend monobromonaphthalene for this purpose. 
In the case of a liquid, it is contained in a glass cell cemented to 
the prism. 

Fig. 170 

Light is directed into the liquid in a direction almost parallel 
to the horizontal surface so that light entering the prism makes the 
critical angle, c, with the normal. 

Let it emerge from the prism at an angle, i. (Cf. fig. 171.) 
Suppose that the refractive index of the substance to be examined 
is n and of the material of the prism, ii - 


sin c = — » 
sin * 

sin (| - c) 




cose = 


sin 8 c -f cos 2 c = 1 = ^- + 

sin 2 i 

t* — V(Jt 2 — sin 2 *' 
The substance must have a refractive index less than that of 
the material of the prism if it is to be examined by this method. 

The apparatus measures the angle, *. For this purpose a 
telescope is attached to a circular scale and the rays are received 
by it. Since no rays within the prism make an angle greater 

Fig. 171 
than c with' the normal on entrance to the prism, the angle, 

\i ~~ c ) ' measures the minimum angle at which rays strike the 

vertical face. Corresponding to this, i measures the minimum 
inclination of the emergent rays to the normal. Thus in the 
telescope the rays emerging in this direction bound the field of view. 

The apparatus is arranged so that the rays are deflected down 
the telescope, and when its cross-wires lie on the dark edge of the 
field the scale reads off the angle, *', to an accuracy of one minute of 
arc. no, of course, is an instrument constant and has the value 174. 

In addition a microscope screw is divided so that the value of 
the refractive index may be measured for different wave-lengths 
to a still higher degree of accuracy. 

The prism and specimen are surrounded by a metal water 
jacket and thermometers are provided for reading the temperature. 
A table is supplied with the instrument giving the value of t* 
corresponding to different values of *. 

Determination of the Focal Lengths of Thin Lenses by means of Pins 
Convex Lens. Method, 1 

Support the lens vertically and place a vertical pin behind it 
so that its point lies on a level with the centre of the lens and on 
the axis of the lens. 


The adjustments are rendered easier by placing a sheet of 
white paper behind the pin. 

Set up another pin so that it coincides with the image. When 
this is the case the image and second pin will not appear to move 
relatively to each other when the eye is moved horizontally from 
side to side. 

Arrange the lens and first pin so that the image is not at the 
same distance from the lens as the object. Then without moving 
either pin displace the lens until a second position is found at 
which the pins occupy conjugate points. 

Fig. 172 

Let the lens have been moved a distance, d cms., while the 
distance between the pins is D cms. 
In the figure we have evidently : 

F^i = u = — — , 

and C^Pa = v = — ^— (numerically). 

Thus from the equation : 
1 _ 1 1 

v u ~~ f 

we have : / — — — ^n — ' 

since in the formula we have to write : 

D +d 

v = — — . 


Determine several different values of D and d, and calculate the 
value of /from each ; the result should give a constant. 

Method 2 

Put up a plane mirror immediately behind the lens and parallel 
to it. Then place a pin in front of the lens as in the first case 
and adjust it to make it coincide with the image formed by 
refraction in the lens and reflection in the mirror. The distance 
from the pin to the lens is equal to the focal length of the lens, 



for the rays are reflected back along their path and must therefore 
strike the mirror normally. They leave the lens las parallel rays, 
and must therefore originate from the principal focus. 

Fig. 173 

Method 3 

Set up a plane mirror, P^ 1 , at some distance from the lens, 
and a pin on the other side of the lens. The pin should be 
mounted so that its centre is on a level with the centre of the lens 
and should be adjusted until the position of the image of the j>in 
is made to coincide with the pin. The way this is brought 
about may be seen from the diagram. The image formed by the 
lens evidently lies on the surface of the mirror. 


Fig. 174 

If the distances be measured we have the positions of a pair of 
conjugate points. 

Make several observations for different distances and calculate/. 

In making experiments with a convex lens it is useful to know 
the focal length approximately before making an accurate 
determination of it. Sometimes time is wasted in trying to locate 
a real image when the object is so placed that a virtual one is 
formed. It should be noted that if the object is at a distance 
from a lens, which is less than the focal length, the image is virtual. 

Focus a distant object — a lamp or window, if not too close, 
is suitable — and measure the distance from the lens to a well- 
defined image thrown on to a sheet of paper. Unless the lens 
has a very long focal length this distance will be approximately/. 
Concave Lens 

Method 1 

Put up the lens and place a convex lens of known focal length 
in contact with it so that their axes coincide. 


Note if the combination will form a real image of a distant 
object, and measure approximately the distance from the lens to 
image so formed. This will indicate whether the combination 
acts as a convex lens. If it does not the convex lens is too weak, 
and a stronger must be chosen. 

Determine the focal length, F, of the combination by any of 
the above methods, and let / denote that of the convex lens, f 1 
that of the concave. 

Calculate f 1 from the formula : 

Fig. 175 

Method 2 

Set up a concave mirror behind the lens and a pin in front. 
An inverted image of the pin will be seen on looking through the 
lens, and it may be made to coincide with the pin by suitably 
adjusting the mirror. 

The diagram illustrates that this is brought about by the normal 
incidence of the rays emerging from the lens on the mirror. 

The virtual image of P is thus at P 1 , the centre of curvature 
of the mirror. 

Remove the lens after noting its position and the distance of P 
from it, and again adjust the pin until it coincides with its image 
formed by the mirror. We thus locate P 1 and can measure v, 
the distance from the position of the lens. 

On substitution in the formula : 

1 __ 1 _ 1 
v u ~J 
we calculate /. 

Repeat this for several different cases. 

The Focal Length of a Lens Combination 

In the case of a thin lens, in which no account of the separation 
of the two surfaces is taken, there are two points on either side 
of it called principal focal points at equal distances from the 
lens if it is situated in a medium which is uniform all round it. 


These two points are distinguished by calling bne the first and 
the other the second focal points. The former is the position 
of an object corresponding to an image at infinity while the 
second is the position of an image corresponding to an object 
at infinity. 

A plane perpendicular to the axis at the position of the lens is 
called the principal plane and if / denote the focal distance, we 
have the relation : 

v u f* 

connecting the distances of the conjugate points from the principal 

If the principal plane and focal points are known it is a matter 
of simple geometrical construction to determine the image of 
any object. 

When the lens is not thin, or when it is necessary to deal with 
a system of lenses, Gauss has shown that the formula is the same 
but that in this case there are two principal planes, the first and 
second, separated by a finite distance. 

The distance of the first principal plane to the first principal 
focus is called the first focal length, and the distance to the second 
principal focus from the second principal plane is the second 
focal length. 

If the lens system is situated in a medium the same on both 
sides the two focal lengths are the same and we shall denote 
either by /. 

When the object is at a distance,- u, from the first principal 
plane and the image is at a distance, v, from the second, the formula 
is still : 


v u ~~~ f* 

A very important property of the principal planes is that if 
a ray is incident on the system towards a point on the first 
principal plane, the emergent ray is directed from a point in the 
second principal plane which lies on the same side of the axis 
and at the same distance from it as the first. 

It is now possible by a geometrical construction to determine 
the position of an image corresponding to any object. This is 
illustrated in the diagram (fig. 176). 

The ray, BI^, parallel to the axis must pass on emergence 
through the second focus, F a , and it appears to emerge from L 2 , 
where PiL x = P 2 L 2 . 

Pj and P 2 denote the principal points and PiL^ and P 2 L 2 are 
the principal planes. 


Again the ray through F 2 emerges parallel to the axis from M 2 > 
where P 1 M 1 = P 2 M 2 . 

We thus find B 1 and can draw in the image, ATO 1 . 

If the media on either side are different, F 1 F 1 is not equal 
to P 2 F 2 . 

Two other important points for the lens system are the nodal 
points. They are defined to be points on the axis such that a 
ray directed to the first nodal point emerges as a parallel ray 
directed from the second. 





V ** 

\F<, A 1 






B 1 

Fig. 176 

When the medium is the same on both sides of the system the 
first and second nodal points coincide with the first and second 
principal points. 

We have, therefore, the line, BP^ parallel to PaB 1 , and the 
magnification, m, is 

A^ 1 = A X P 2 = v 



just as in the case of a thin lens. 

A direct determination of / requires the location of a pair of 
conjugate points, but we do not know the position of the principal 
points so that it is not possible to find v and u for substitution 
in the formula : 

v u f 

We may, however, determine / by measuring the magnification 
of an object by the lens for two different positions. 
From this formula we have : 

— = 1 

Thus if u x and m x denote the distance of an object from the 
first principal plane and the corresponding magnification, we have : 

m x f 



and for a second position : 



/. / = 


I I 
m x m z 
Thus the actual lengths, u, are not required for the calculation, 
but only the distance between the two positions of the object, 
which is given by the difference between u x and u 2 . 

If several values of — are observed and positions corresponding 

to u noted on the optical bench, a graph may be plotted of — 
against u. 

Fig. 177 

The relation is linear, and if two points, P t and P 2 , be chosen 
on the line as far apart as is convenient : 

P 2 N 2 - P X N X 

NoN, < 

It is convenient to use a transparent glass scale as object and 
to illuminate it by monochromatic light. A similar glass scale 
is used as a screen so that the apparent length of a certain number 
of divisions of the first scale seen in the image can be read off at 
once on the second scale. 

The screen is adjusted as usual by noting when there is no 
parallax between the image and scale. 

For accurate work it is necessary to use a low-power microscope 
to measure the image. 

Another method which is instructive is to adjust the positions 
to obtain a magnification of magnitudes 1 and 2. 


When the object and image are equal in size their distances 
from the corresponding nodal points are both 2/. 

This may be verified from a diagram or from the formula. 
Now keep the screen fixed and move the object and lens until 
the magnification is 2. 

It may be verified that in this case the object is — / from the 

first nodal point, and the image is 3/ from the second. The lens 
system has thus been moved through a distance, /, and this may 
be measured directly. 

Suppose the diagram (fig. 176), denotes the case for a magnifi- 
cation unity. 

In this case AP X and A X P 2 are both equal to 2/. We know the 
value of /, so that since A is a definite known point we can locate 
Pj and similarly P 2 is known from the point, A 1 . 

The lens system is sometimes encased in a tube, and in that 
case the distances from P x and P 2 to the ends of the tube should 
be recorded. 

If the lens is merely a thick lens of glass the distances of the 
principal points from the faces of the lens should be recorded. 

Determination of the Principal Planes of a Thick Lens 

Let DAEB denote a thick lens and let C and C 1 be the centres 
of curvature of the faces, DBE and DAE respectively. 

There exists a point, O, such that all the rays which pass 
through it emerge from the lens parallel to their original direction. 
This point lies on the axis, CC 1 , and is called the centre of the 

The rays passing through O after their first refraction, were 
directed originally towards a point, P. Thus P and O are 
conjugate points with respect to the first surface. 


The rays after the second refraction appear to come from P 1 , 
so that P 1 and O are conjugate points for the second surface. 
These points, O, P and P 1 , are fixed points for the lens and P and 
P 1 are the principal points. 

If the distance of an object is measured from P and denoted 
by u, while the image is measured from P 1 and denoted by v, 
we have the relation : 

1 _£ _ 1 
v u ~ f' 

as in the last experiment. 

The quantities, v and u, are to be given positive signs if measure- 
ment is made towards the source of light, and negative if it is 
made in the opposite direction. 

Parallel light incident on the lens on the right is brought to a 
focus at F 1 where PT 1 =/. 

Light originating at F is parallel on emergence where PF = — /. 

Let a source of light be located at a point, S, where FS = x, 
x is to be regarded as a number of cms., and is not given a sign. 
The focal distance is likewise measured by the number F. The 
lens, when convex, has a negative focal length, i.e. in the formula 
/ = - F. 

The image will be formed at I, say at distance, y, from F 1 . 

Then u = (F + x), and v = - (F + y). 


^ » 

F +y F + x F 

or X y = F 2 . 

The disposition of the conjugate points shown in the diagram 
is such as occurs in the case of a convex lens. The object and 
image both lie within the focal distance or without it. In this 
case x is taken positive when measured from F towards the right, 
and y is positive when measured from F 1 towards the left. This 
relation enables us to find F, for although it is not easy to locate 
P and P 1 , which are required if the first formula is used, it is easy 
to locate F and F 1 , and hence to measure the distance of object 
and image from them. 

To locate F, set up a plane mirror on the left of the lens and 
put up a pin in such a position that its image coincides with it. 
The rays emerging from the lens must in this case be parallel 
and strike the mirror normally so that object and image lie at F. 

Repeat this for the point, F 1 , leaving the former pin in position. 

Leave pins at F and F 1 and do not move the lens. Place a 
third pin at a point such as S, farther from the lens than F, and 
locate its image, I. 


The distances, x and y, are now determined and we can calcu- 
late F. 

Hence since P and P 1 lie at distance, F, from F and F 1 respec- 
tively, the principal points are located. 

Repeat this for several values of x and y. 

"Piano-Convex Lens 

In this case any ray, R, striking the lens at P in the figure, 
where P is the pohit of intersection of the axis and curved surface, 
is not deviated but gives rise to the parallel emergent ray, R 1 , 
for it is just as if refraction took place in a slab of glass bounded 
by the face, A, and the tangent at P. "R 1 appears to Come from 
P l , so that P and P 1 are the principal points. These may be 
located as above. 

Fig. 179 

We may, also, calculate the distance AP 1 . 

For AP 1 is the apparent thickness of the widest portion of the 


AP _ t 

•'• '""AP 1 d 

This gives d if \l and t are known. 

It forms a useful exercise to determine P and P 1 by the method 
described above, and then to deduce y. from the observations. 
If PP 1 = 5, 


* = *— *• 

Determination of the Principal Points by Rotation of the Lens 

If the lens is mounted on a stand which can be rotated about 
a vertical axis we may use the property of the nodal points to 
locate P and P'. 

It is convenient to place the lens in a holder which can slide 
along a scale fixed to the rotating stand. The arrangement of 
the apparatus is illustrated in fig. 180. 

A mirror is fixed on one side of the lens normally to the axis 
and does not rotate with the lens. 



On the other side a pin is set up and moved until its image, 
by two refractions through the lens and a normal reflection at 
the mirror, coincides with the object. 

The pin then lies at a principal focus. 

The lens is moved along the scale and the pin adjusted so that 
image and object coincide until a position is found, when slight 
rotations of the stand fail to cause displacement between the 
pin and its image. When this is the case the rotation takes place 
about a vertical axis through the nodal point nearer the pin. 

The figure explains this, for if P' denote the other nodal point 
in the symmetrical position and P" the position of this nodal 
point in a slightly displaced position a ray from F falling on P, 
first passes along PP'M and is reflected back along its path after 
incidence on the mirror at M. 

Fig. 180 

When the lens is rotated into the position indicated by the 
dotted lines the ray, FP, is refracted and emerges from the lens 
parallel to its original direction, and directed from P', i.e. it takes 
the course, P'M', and is reflected back to P', emerging once more 
along PF. The image of the point therefore keeps the same 

The position of the axis thus fixes the nodal point, and the 
axis is usually clearly indicated on the stand. This finds the 
point, P, and, by turning the lens, P' may be found in the same way. 
The Measurement of the Focal Length of an Optical System by means 
of a Goniometer (Searle's Method). (Proc. Optical " Con- 
vention," 1912, p. 165.) 

A simple form of goniometer devised by Dr. G. F. C. Searle, 
in conjunction with Messrs. W. G. Pye and Co., provides a*very 
instructive method of determining the focal length of an optical 
system by means of the properties of the nodal points. 

The goniometer is illustrated in fig. 181, and consists of a 
wooden base provided at one end, with a spherical pivot, con- 
sisting of a ball of phosphor-bronze, and carrying at the other a 
scale marked in millimetres. 

A movable arm rotating about the pivot carries an achro- 
matic lens of focal length about 35 cms., a vertical adjustable 
frame, across which a vertical wire is tightly stretched, and a fine 
wire passing across an opening which serves as a scale index. 


The ball is adjusted at a distance of 40 cms. from the scale, 
which can be read to an accuracy of one-tenth of a millimetre, 
so that a rotation of the arm through one-seventieth of a degree 

is able to be measured. This is about —5- radian. 


Fig. 181 

In fig. 182, let HiB and H^B 1 denote the principal planes of 
an optical system, the points, H^ and H 2 , denoting the first and 
second principal points. 

In practice it usually happens that the medium on each side 
of the system is the same, viz. air. As the general case presents 
no additional difficulties and can be examined also by this method, 
we shall write jx x for the refractive index of the medium on the 
right of the system and i* 3 that of the medium on the left. Then, 
if U is measured from H t and V from H 2 we have the relation, 

V U F 

between the object distance, U, and the image distance, V. 
F is a constant for the system, and if the value of U is F x for 
emergent parallel light, i.e. Fj denotes the first focal distance, 

Fi = - i*iF, 
and similarly the second focal distance, F 2 , is given by : 

F 2 = [x 3 F. 

The usual sign convention is adopted in applying these formulae, 
directions from the lens towards the source are positive and those 
in the opposite direction are negative. In fig. 182, F x and F 2 
denote the first and second focal points respectively. 

Nj and N 2 are the nodal points, and with the same medium 
on either side of the system they coincide respectively with H^ 
and H 2 . 

In the general case the first nodal point lies at the distance 
(F x + F 2 ) from K v and N 2 is at the same distance from H 2 . 




and similarly : 

NJF, = H.Fj - H 1 N 1 

= F 1 -(F 1 + F 2 ) = -F 2 , 

N 2 F 2 = H 2 F 2 + H 2 N 2 = F v 
As drawn in the diagram the value, F t = F^, is positive, 

since it is measured from H lf while F 2 is negative. 

Any ray incident on the system in a direction passing through 

N x gives rise to an emergent ray in a parallel direction passing 

through N 2 . 

Fig. 1 8a 

Thus the ray, ANj, gives rise to CD, and CD if produced would 
pass through N 2 . 

If the point, A, lies in the focal plane a divergent pencil from 
it gives rise to a parallel beam. 

Two rays, AB and ANj, are drawn and the corresponding ravs 
are BT. and CI). 

Since CD is parallel to AN X the angle between the axis and the 
emergent beam is equal to angle, ANjF v 

Thus if we measure the distance, AF t , denoting it by d, we can 
find the length, N^, provided that we measure the wangle 
between the axis and the parallel beam. 

If this is denoted by a we have : 

NjFi = dcot a; 

or when a is small, N^ = - = F 2 (numerically). 

The goniometer is a very convenient apparatus for carrying 
out this measurement. 

It is first adjusted so that the vertical wire lies in the focal 
plane of the lens. This may be done by adjusting the wire so 
that the image of a distant object seen through the lens falls on 
the wire; or a plane mirror may be placed near the lens on the 
side opposite to that on which the wire lies and the latter adjusted 


until there is no parallax between the wire itself and its iihage. 
The frame is then clamped. 

The optical system is now brought up so that its axis lies 
collinear with that of the goniometer lens when this lens lies so 
that the horizontal wire index is central. 

A glass scale is placed in the focal plane of the optical system 
and may be represented by AF r A denotes one of the marks 
on the scale. The scale may be adjusted in the focal plane by 
looking through the goniometer lens and the system and by 
placing it so that its image is seen coincident with the vertical 
goniometer wire. The method of parallax is used to obtain an 
accurate adjustment. When this is made as closely as possible 
by means of the unaided eye a lens should be used as a magnifying 
glass to make it still more exact. 

Note the reading on the glass scale when the apparatus is in 
the symmetrical position, so that the point corresponding to Fj 
is observed. Note also the indication of the goniometer scale. 

Turn the goniometer through some convenient small angle 
and note the mark of the scale seen coincident with the vertical 
wire. Again note the goniometer reading. We thus have the 
values of AF X and of the angle, a, and can consequently deduce 
F 2 . By turning the system end for end, so that H 2 lies to the 
right of Hj, we find F x . 

We can thus locate the positions of N x and N 2 with respect 
to the outer surfaces of the system. These should be recorded, 
and since HjNj = F! + F 2 = H 2 N 2 we may also locate the 
principal points. Record these also. 

In order to carry out this experiment it is convenient to employ 
two thin lenses situated at a known distance apart. These form 
an optical system of the type described, and having the same 
medium on each side. 

Hence N x and N 2 are both nodal and principal points, and F! 
and F 2 are numerically equal but of opposite sign. 

Find the focal length and the position of the principal points 
for this case. 

It is shown in treatises on Optics (see, for example, Houstoun's 
"Treatise on Light," p. 45), that for two lenses separated by a 
distance, d, the position of the first nodal point is at a distance 
from the lens on which light is incident and whose focal length 
is/i, given by 

h+h + d> 
while the second nodal point is at a distance : 




from the second lens, f a , denoting its focal length, while the focal 
length of the system is : 

F = A/e . 
fx+f % + * 
Measure f 1 ,f 2 , and d, and verify these results, 
/i aim / 2 may be readily measured by the goniometer by re- 
placing the optical system by the lenses in turn. 

Another property of this system, which is of importance in the 
theory of Optics, concerns the lens equivalent to this system. 
This lens is such that it produces an image of the same size, but 
not in the same position, as that produced by the system, when 

situated at a point at a distance from the first lens equal to — -?-, 

i.e. at the first nodal point. 

Fig. 182, shows the relative position of the nodal points in the 
case of two convex lenses. 

TJie Focal Length of a Microscope Objective 

The goniometer provides a useful means of measuring a short 
focal length, such as that of a microscope objective*. 

The screen, AF 1} is replaced by a micrometer slide provided 
with a magnifying reading glass. The slide is provided with a 
small engraved scale with divisions at each tenth of a millimetre. 
The objective is placed to receive parallel rays emerging from the 
goniometer lens and focusses them on the scale. Thus an image 
of the goniometer vertical wire is received and is viewed by the 
magnifying glass. 

The position is read for the symmetrical position and again 
after a slight displacement of the goniometer arm. The image 
is now focussed at a new point corresponding to A (fig. 182). 
The distance between the two images is read off, and from a know- 
ledge of the angle of rotation of the arm, the focal length is 
calculated as before. 

The Determination of the Foeal Length and Principal Points of an 
Optical System by means of the Revolving Table 

The principle of the method is the same as that described on 
page 302, for it depends on the same property of the nodal points. 
The revolving table affords, however, a much more convenient 
and accurate means of carrying out the experiment. (See figs. 
152 and 183.) 

Each nodal point is determined by the principle that when a 
small rotation is made about a vertical axis through the second 
nodal point, the image produced of a distant object is not dis- 
placed. Thus the position of the axis of rotation of the table 
locates the second nodal point. 


The lens system is mounted on a slider with straight edges, 
and is placed on the table so that the edge slides along the scale, 
An index mark on the slider serves to locate positions with respect 
to the scale. 

The design secures that there is no side shaking. 

It is essential in this experiment as in the former (p. 268), to 
make the axis of the system intersect the axis of revolution. 

Fig. 183 

In order to arrange this the image of a distant object is made to 
coincide with a pin held in a clamp and the system is moved on 
the table until, with this focussing exactly made, a slight rotation 
of the table causes no displacement between the image and pin. 

Suppose that the axis of the system is represented by N X N 2 
(fig. 184), and that S denotes the distant source, while I is the 
image, S is so far distant that it may be supposed to lie on either 
of the lines, IN lf N^L, I 1 N V 

I Nt_N, s 

I ! L ZZT. * 

A n! l 


n, n z 

Fig. 184 

Let A denote the point of intersection of the axis of revolution 
and the horizontal plane. 

If AN 2 is normal to the direction in which the light is travelling 
a slight rotation does not displace N a at right angles to the axis. 
Thus the light incident at the first nodal point still emerges along 
N a I, and I is not displaced. 


Nj 1 denotes the displaced position of N x and the dotted line 
LNj 1 denotes an incident ray from S. 

Thus with this setting the nodal point, N a , lies in a plane through 
the axis of revolution and normal to the direction of the rays. 

The table top is now turned through two right angles, and the 
lens system moved along the scale until the distant object is 
focussed on a second pin as before. 

In general this pin will not coincide with the first, but will be 
displaced in a direction normal to the direction of the light. 
This is illustrated in the figure. Nj in the second case become 
the second nodal point. 

If I 1 denotes the second pin it is clear that the axis lies on a line 
drawn parallel to the light and passing midway between the pins. 

The scale on the table top is undamped and moved through 
half the distance between I and I*, so as to carry the axis of the 
system into the correct position. 

The apparatus is now once more carefully adjusted so that 
the image does not move for slight rotations. 

The axis then passes through the nodal point. The system is 
then turned end for end and the second nodal point found. 

In order to find the focal length readily, a clamp is fixed to the 
table which carries a scale (fig. 183). 

The pin is left in position and the table turned until the scale 
just touches it and the scale reading is noted. The table is turned 
so that the other end of the scale just touches the pin and the 
scale reading is noted for this case also. The distance between 
the two marks on the scale is determined by subtraction ; one- 
half of this gives the focal length. 

A diagram is then made showing to a convenient scale the 
positions of the cardinal points with respect to the first and final 
surfaces of the optical system. 

The apparatus may be employed to examine such a system as 
that described in the previous experiment with the goniometer. 
In this case the apparatus is not very convenient for the measure- 
ment of short focal lengths. It is better to use the goniometer 
for this purpose. 

When a distant object is not available in the above experiment 
it is only necessary to place a plane mirror on the side of the lens 
system away from the pin and adjust the pin until image and 
object coincide. 

The focal length of a thick lens, the radii of the surfaces being 
r and r 1 , and the thickness t, is given by : 

F== ¥* 

>-!).{ nfr 1 - r) - (n - 1) t\ ' 

Here r and r 1 must be given their proper signs. 


It is instructive to determine the value of the refractive index 
of the glass forming a thick convex lens. 

Determine r and r 1 by the revolving table method (p. 269), 
and F by the method just described, while t may be measured 
by means of callipers. 

Solve the resulting equation for i*. 

The experiment is intended to provide practice in working with 
a thick lens and becoming familiar with the formula for F. 

The distance between the principal points is given by 
t Qi - 1) (r 1 - r - *) , 
(x (r 1 - r) - ((i - 1) t 

This should be verified by the results of the experiment which 
locates the position of the nodal points, which are the same as 
the principal points in this case. 

The student should verify the two expressions given above. 

Spherical Aberration with a Thick Lens 

In the simple theory of lenses it is assumed that all the rays 
starting out from a point on the axis are brought to a focus after 
refraction through the lens to another point on the axis. 


Fig. 185 

This is only approximately true for rays lying very close to 
the axis. A ray, PA X (fig. 185), will be made to cross the axis 
at F, and if PA X makes a small angle with the axis, F will be the 
focus for all such rays. But a ray, PA 2 , will cut the axis at C 2 , 
and PA 3 at C 3 , after refraction, these points lying closer to the lens 
as the incident ray is more inclined to the axis. 

The distances, FC 2 and FC 3 , are called the longitudinal aberra- 
tions of the corresponding rays. 

The distance, FC 2 , depends on the inclination of the emergent 
rays to the axis. If this distance is x and the tangent of the 
acute angle at C 2 is m, we may say that x is some function of m. 
We do not know the way in which x depends on m, but whatever 
be the form of this dependence we may write x in powers of m, or 

x = a + bm + cm 2 + dm 9 + . . . , 
where a, b, c, d, etc., are constants. 


Since the lens is symmetrical about the axis, FC 2 is the same 
for positive and negative values of m. 

Thus if we write — m instead of m in the above equation we 
have the same value of x, so that x depends only on even powers 
of m. 

.'. x = a + cm 2 + em 4 -f . . . 

When the emergent ray has a very small value of m, the' point, 
C 2 , is at F, or x vanishes. Thus a vanishes and we have : 
x = cm 2 -f em 4 -f . . . 

If m is not very big, m 4 and higher powers of m, will be very 
small, so that we may write : 

x = cm 2 . 

Let F be taken as origin and FP as axis of x. 
Let v denote distances measured at right angles to FP. 
The line, B 4 C a , has an inclination to the x axis whose tangent 
is m, and it passes through the point, C 2 , with co-ordinates (cm 2 , o). 
So that its equation is : 

y = m (x — cm 2 ). 

The emergent rays cross one another at points B 2 , B 3 , B 4 , etc., 
and the points at which consecutive rays cross lie on a curve 
known as the caustic of the lens. 

Suppose that to a ray consecutive to that through C 2 
the corresponding tangent is m 1 . 

Its equation will be : 

y = m 1 (x — cm 12 )- 

If we solve these two equations we can find the co-ordinates 
(x, y) of the point common to both lines. We do not actually 
require these but only the relation between these two co-ordinates, 
which will give a locus of the intersections of consecutive rays 
or the equation to the caustic. 

Subtracting one equation from the other : 

x (m 1 —tn)=c (m l3 — m 3 ), 

x = c (m 12 + mm 1 -f m 2 ) 
= 3cm 2 , 

since m and m^are so nearly equal ; 


'3c)» C ' \3C J ~ 3 ' (3cY 

(3c)* \3cJ 3 ' (3c)* * 

or 2j,cy 2 = 4# 3 . 


Thus for values of m, such that m* is negligible the form of 
1 the caustic is a semi-cubical parabola. 

In the figure, DCE 1 and D*CE are the extreme rays from the 
lens and DEF, D X E X F the two branches of the caustic curve. 

Fig. i 88 

All rays inside the extreme rays touch this curve at points 
which are nearer to F, the smaller the inclinations of the rays to 
the axis. The rays are consequently crowded together just 
inside the caustic curve, and a screen placed between C and DD 1 
will show a circular patch of light with a brighter illumination 
round the circumference. At points between C and EE 1 , a 
brighter patch appears in the centre which increases to EE 1 where 
the illuminated area is a minimum. EE 1 is called the circle of 
least aberration. 

Beyond EE 1 the first circle diminishes until at F it becomes a 

Set up a large lens on the optical bench with a small hole 
illuminated by a Bunsen flame containing common salt as a source 
of light. The light may be concentrated on the hole by means 
of a small condensing lens. 

It is best to have the distance from the hole to the large lens 
about twice the focal length of the lens. 

Examine the image by means of a micrometer eyepiece, when 
the appearances described will be seen if the source lies on the 
axis of the lens. 

Find the position F, and read off the position of the eyepiece 
on the bench scale. Move in towards the lens and measure the 
diameters of the circular rings at a series of distances from F. 
Plot the results on squared paper and verify the equation of the 
caustic curve. 

The radius of the circle of least aberration is r = J cm 3 , 
where c is constant in the equation and m Q is the value of m for 
the extreme rays. 


tn is nearly proportional to the aperture of the lens. Vary 
the aperture by putting diaphragms over the lens and measure 
the corresponding diameters of the least circles. Verify that the 
diameter is proportional to the cube of the aperture. 

The above value of r is readily calculated by remembering that 
it is the value of the ordinate of the caustic at the point where 
it is cut by the extreme ray. 

It is therefore necessary to solve the equations : 

v = m (x — cm 2 ) {4) 

2ycy 2 = 4a; 3 . 

The solutions are 

x = ^cm^, y = ± I cm *. 

In order to find the value of FL, we note that by equation (4), 
which is the equation of the line, CL, the value of v corres- 
ponding to x = o is — cm 3 . 

This negative sign occurs because we have measured the angle 
of which m is the tangent from the x axis in a clockwise direction. 

In the case of the line, CL, the value of m is negative since the 
angle is obtuse. 

The lateral aberrations of the extreme ray, i.e. the length of 
FL, is thus — cm *. 

But the radius of the circle of least aberration is £cw 3 (nu- 

Thus the ratio 7^=; = 4. 


Verify this result. 



The theory that light consists of waves in the ether leads us to 
expect, by the principles of superposition, that class of phenomena 
described by the term 'interference.' Particles of a medium, 
when simultaneously displaced by the arrival of several disturb- 
ances, have a resultant displacement obtained by adding together 
the vectors representing individual displacements. In particular, if 
a particle is subject to displacements in directly opposite directions, 
it will be displaced a smaller amount than if it were subject to 
either separately, and if both the displacements are equal, but 
oppositely directed, there will be no displacement of the particle. 
On the other hand, displacements arriving at a particle in the 
same direction will cause it to be moved a distance equal to the 
sum of the separate displacements. 

The arrival of oscillatory disturbances at any point may, 
therefore, cause larger or smaller displacements of an ether 
particle than would occur as a result of each separately. 

The intensity of illumination at any point of a medium is 
proportional to the square of the amplitude of the vibrations 
executed by the ether particle at that point. 

It can thus be seen how it is possible to produce in the ether 
places of large or small intensity as a result of the arrival of 
two trains of waves. It may even be possible that there will 

Fig. 187 

be darkness at certain points since it is possible that as a conse- 
quence of adding vectorially the displacements we get no net effect. 

This point will be considered in detail. 

Let Sj and S 2 denote two sources of monochromatic light. 
From each is thus emitted a train of waves of a particular wave 



length, the same for each. In addition, suppose that the distur- 
bance starting from S x is in the same phase as that starting 
from S 2 . By this we mean that the ether particles at S x and S a 
are in exactly the same state of vibration, they are moving in 
the same direction, and are displaced the same amount from 
their central positions. These disturbances are propagated with 
a particular velocity in the ether, and at points equidistant from 
S t and S 2 the displacements will be in the same phase when they 
arrive there. In other cases where points are at different dis- 
tances from S x and S 2 , it may happen that the displacements 
are in different directions on reaching the point. Draw AB at 
right angles to the middle point of S^g and let it cut a screen, 
MjMa, normally at B. 

We shall inquire as to the illumination at different points on 
the screen. The disturbances will reach B in the same phase, 
the displacements will be in the same direction, and will thus 
unite in increasing the illumination at B, which will always be 

Of course, we think of the waves as transverse, i.e. the motion 
of the particles is normal to the direction of propagation. 

Take any other point, P, distant x n from B. 

The light from S 2 travels over the path, S 2 P, and that from 
S x over SjP. Thus those disturbances, which reach P simul- 
taneously, started at different times from their sources. 

The vibrations will thus be in different phases on reaching P. 
If, however, the difference between S 2 P and S X P is a whole 
number of wave lengths, the disturbance from S a left that point 
a complete number of periods before the disturbance from S x 
set out towards P. The displacements are thus in the same 
direction, and will unite at P to give brightness, just as they do 
at B. 

If the difference of path is an odd number of half wave-lengths, 
the displacements on reaching P will be directed oppositely, and 
will tend to destroy the disturbance at P. If the amplitudes 
are equal there will be no displacement, and consequently no 
illumination at P. In any case there will be a marked falling 
off in brightness. We ought, therefore, to be able to distinguish 
on either side of B as we proceed outwards alternate bright and 
dark places. 

Suppose the distance AS! = AS 2 = d, while AB = D. 
Then PMj = PB -MjB 

= x n — d, 
and similarly PM 2 = x n -f- d. 

(Mi and M 2 are the feet of the perpendiculars from S x and S 2 on to 
the screen.) 


.♦.S 2 P={D* + (*„+<*)*}* =D|i+^^j (i) 

x -\-d 
in which we neglect higher powers of -—■ — than the square. 

This means that we suppose that x n and d are of small magni- 
tude relatively to D. This will evidently be the case if we 
confine our attention to a few only of the alternations in intensity 
about B, for the distances apart of adjacent bright points on 
the screen are of a magnitude comparable with the wave lengths 
of the light, and this is very small. 

In the same way 

S t P = D{l+i2iL^)!j; ( 2 ) 

Thus for a bright point : 

and for darkness 

D x n = «x, (3) 

2d 2W + I ^ , , 

D*« = ^-- X; * (4) 

or writing 8 = 2d, i.e. SjSg = <5, 

we have x n • yz= «xor(w + |)x, (5) 

according as P is bright or dark. 

In the experiments to be described (5) is of fundamental 

The first three experiments are three examples of obtaining 
sources S 2 and S 2 of the kind described. 

It would be useless to set up two slits and illuminate them by 
a sodium flame, for we have in such a flame a multitude of 
sources in different phases. 

The method adopted is to cause light from a single source to 
travel to the points of which P is typical by two different routes, 
and to unite on arrival. We thus have the equivalent of two 
sources emitting vibrations of the same phase. 

In making experiments on interference and diffraction accurate 
measurements have to be made ; for this reason the optical 
bench is used. It consists of a strong, rigid metal frame, provided 
with levelling screws on which it stands. The frame consists 
of two metal rails, one of which is graduated accurately in 
millimetres. Along the rails slide metal uprights, each of which 
is attached to a vernier at its lower end so that its position on 
the bench may be accurately read off. 


The uprights serve to carry a slit, micrometer, microscope, 
lens, or whatever piece of apparatus is necessary. If it is 
necessary to move any piece of apparatus transversely across 
the bench, it is placed in an upright fitted in a support provided 
with a transverse micrometer screw. 

Fig. 188 

Although it is possible to read off positions of the uprights on 
the bed of the frame, since the slit or cross-wire may not be 
exactly above the indicator mark we cannot read off directly 
the distance from slit to cross-wire. A correction must always 
be applied. In order to find this correction a stand, carrying 
a carefully measured rod is placed on the rails, one end of the 
rod is placed in contact with the slit while the other end is viewed 
in the micrometer. Let the length of the rod be I, and suppose 
the distance between the slit and micrometer uprights is I 1 , as 
observed on the scale. Then to convert the readings as obtained 
from the upright to the distances required we must add to the 
observed readings the quantity (I — I 1 ). 

Determination of the Wave Length of Sodium Light by means of 
Lloyd's Single Mirror 

The simplest way of obtaining interference bands is by means 
of a mirror silvered on the front surface or blackened at the 
back in order to avoid multiple reflections. 

The diagram illustrates how the interference is brought about. 

Light from a slit, S lt travels directly to a screen, PB, and 
also by the alternative path after reflection at the mirror, MM. 
For example, a ray may reach P by the direct path, SiP, or by 
the path, SjCP. The latter ray produces the same effect at P 


as would arise from a ray, S 2 P, S 2 denoting the position of the 
image of S x in the mirror. We thus have the equivalent of two 
sources, S x and S a , emitting vibrations in the same phase, except 
for any change of phase that may occur on reflection at MM. 
MM is mounted vertically on an upright of an optical bench, 
and the slit, S r , is carried vertically on another upright. 

1 P 

- B 

S» Fig. 189 

MM is mounted as accurately parallel to S x as possible, and the 
fringes are looked for by the micrometer microscope. They will 
usually come into view, if not already there, after slightly rotating 
the mirror, and are rendered distinct by adjusting the width 
of the slit. 

The mirror should be parallel to the length of the bench, and 
we shall see how to make this adjustment accurately shortly. 

When the fringes appear as definite and bright as possible, 
it is necessary before proceeding further to make sure that the 
distance, SjSg, can be found. 

The method adopted is to use a lens and obtain the magnified 
and diminished images of S^Sg, but we require a lens whose focal 
length is less than a quarter of the distance between S X S 2 and PB. 

In Chapter II, p. 293, we described a method of finding the 
focal length of a lens in which the object and screen are kept 
fixed while the lens is moved. In the diagram in connexion 
with this experiment, suppose that Pj denotes the position of 
the object, S X S 2 , and lt one of the positions of the lens, while 
P 2 denotes the position of the image of S a S 2 . 

By moving the lens to 2 , the image is formed at the same 
place, but it now differs from the former in magnitude. 

Let d lt d 2 denote the respective distances between the lines 
S x and S 2 in the image, while 8 is the actual distance, S X S 2 , between 
the slits. 

Then li_L2l ^! _££_*. 

lhen 8 ~ P^' 8 - OtPj 

But P 1 1 = P 2 2 and P^ = PaO^ 

.-. 8* = d x d z (6) 

Thus, by measuring d x and d z we may deduce <5. 


If MM is too close to PB it may be impossible to obtain both 
images with the lens between the mirror and screen. The mirror 
must be placed so that this is possible. This is the reason for 
performing this part of the experiment first since it is useless 
to measure the distance between consecutive fringes unless we 
can find the distance, S^. The distance apart of the fringes 
does not depend on the position of the mirror, but if the mirror 
has to be replaced in another position it is easy to throw out its 
adjustment. Both images of Sj and S 2 will be in focus in the 
plane of the cross-wires only provided that MM is normal to 
this plane, i.e. to PB. 

It is important that MM should be normal to PB. We have, 
therefore, to rotate MM slightly until the two are accurately in 
focus, the lens being adjusted so that its centre lies on a level 
with the centre of Sj and opposite the edge of the mirror. 

On removing the lens, MM should be rotated slightly about a 
horizontal axis to get the position where the fringes are brightest, 
when S x and MM are parallel. 

Set the cross-wire accurately down the Centre of the first bright 
fringe, and move it always m one direction by means of the 
screws giving the transverse motion, stopping at every three or 
four fringes to note the position. From the observations deduce 
the distance between consecutive fringes. 

Finally measure the distance from S^ to PB. 

From (5) we have : 

N x n = n\ -, 

and proceeding to the next bright band : 


*«+i = (« + 1) x- ; 

.*. % n +l x n = ~7~ (o) 

Let s denote the distance between consecutive fringes, and 
we then have : 

x = 

p y/d x d z . 

As a rule it will be best to make use of the full length of the 
bench, so that D is large, also <5 should be as small as convenient. 
At the same time d x or d 2 must not be too small or it will be 
difficult to make the determination of 6 accurately. 

In recording the results of this and the two following experi- 
ments it is a good plan to begin, as mentioned, at one edge of 
the field, and make out a table as shown below. 


We can then readily obtain a series of independent readings 
for the determination of the separation of the bands. 

In this experiment the first fringe does not lie at the point 
corresponding to B, fig. 187. This is because the reflected ray 
undergoes an abrupt change of phase on reflection at the mirror. 
This has the effect of displacing all the fringes towards B by a 
certain distance leaving the separation between successive 
fringes unaltered. 









(b) - (a) 









lyiean Separation for 15 


Interference by means of FresnePs Double Mirrors 

Interference fringes may also be formed by the use of two 
mirrors silvered on the front surface or blackened at the back, 
very slightly inclined to one another. 

The diagram illustrates the arrangement. 0M X and 0M t 
denote the two mirrors which may be mounted on the optical 
bench and placed accurately vertical. 

S denotes the source of illumination and may most conveniently 
be a Bunsen flame coloured by a sodium salt. 

The fringes are again observed by means of a micrometer, 
and AB denotes the position of its cross-wires. 

The screen, CD, protects AB from direct light from S. 

The source is placed on one side of the bench as the diagram 
indicates, and may be the slit of the optical bench supported 
in a clamp. Sj and S 2 are the two virtual images in the mirrors. 
SxSg and S lie on a circle with centre at O. 


The calculation, mode of measurement, and adjustment, are 
identical with those of the previous experiment. 

The shaded area of fig. 190 shows where rays from both mirrors 
interfere, and the section of this region with the plane of the cross- 
wires of the microscope is the position within which the fringes lie. 


n/ ^ ^~rr rTT7Trr7r777ti 


Fig. 190 
Fresnel's Biprism 

The biprism is a prism with one of its angles only a little less 
than two right angles, and with two equal small base angles. 
The figure illustrates its action. The biprism is represented by 
CDEF, and it acts like two prisms placed base to base. 

Rays from a slit, S, are deviated in each part of the prism and 
unite on the screen, as in the case of the rays, SGP and SHP. 
We have virtually two sources, S x and S 2 . 

Fig. 191 

The details of the experiment are very similar to those described 
in the experiment with Lloyd's mirror. 

The biprism is held in a stand which can rotate about a hori- 
zontal axis parallel to the central line of the bench. By this 
means the edge, D, of the prism can be rotated to bring it parallel 
with the slit. 

The adjustments are first made roughly by the eye, and usually 
the fringes will be observed even with the rough adjustment. 
Slight rotation of the biprism will, as a rule, improve the appear- 


ance of the bands, and still better definition will be obtained by 
narrowing the slit. 

The distance, S^, is measured with the aid of a lens as before. 

In adjusting D parallel to S the following device is of great 
assistance. If the eye is placed on the side of the prism away 
from the slit, and moved across the bench, the slit will appear 
to cross from one side of the prism to the other. As it crosses 
D, unless the slit is parallel to the edge of the prism, the top or 
bottom will cross the edge first, while if parallel it will appear 
to make the transition suddenly. The prism must be rotated 
until this sudden jump occurs. 

Be careful after arranging the apparatus to give good fringes, 
to make the determination of the length, S l S 2 > before measuring 
the separation of the fringes. 

The Determination of the Radius of Curvature of the Face of a 
Convex Lens by means of Newton's Rings 

Newton's Rings are formed as a result of interference between 
the incident and reflected rays from a source of monochromatic 
light on the air film between a plane glass plate and a convex 
lens in contact with it. 


A 1 A 

The diagram shows the lens, L, and the plate, P, in contact 
at the point, Q. 

Two incident rays are drawn, AB and A 1 B 1 . The first is 
partially reflected and refracted at the points, B, C, D, E, and F. 


A ray is drawn which is represented as suffering refraction, 
except at D, where it is reflected. The neighbouring ray, A 1 B 1 , 
which follows the course, A 1 B 1 EFG is shown, and these two rays 
being brought together along EFG will interfere and may produce 
greater or less illumination than each separately, according to 
their phase difference. In calculating this it is to be remembered 
that on account of the reflection at D, at a medium optically 
denser than that in which the ray travels before reflection, a 
phase change of half a wave length is imparted to it. 

The total phase difference between the two united rays is thus : 
(MD + DE + Jx) as reckoned in path length ; for had the rays 
both left the lens, the wave front would have been EM, so that 
the phases are the same at E and M. 

Each ray, such as AB, has a corresponding ray, such as AjB^ 
arising from the same point in the source of light with which it 
can interfere, so that an extended source may be used and a 
Bunsen flame containing sodium acts very well ; in fact, an 
extended source is necessary in order to obtain a large area 
containing rings. 

The rays entering the eye or an optical instrument will be 
contained in a small cone, and for the rays in this pencil the 
change in path difference is very small, so that all the rays 
from the small region of the lens will be naturally reinforced or 
caused to interfere. 

There will be brightness or darkness according as : 
MD -f DE + |x = »x or (n + £)x, 
where n is a whole number, i.e. according as MD -f- DE = an 
odd or even number of half -waves. 

Let CD fall on the plate at incidence 6. 

Let t be the thickness of the film at this point. 

The diagram is drawn with t large, or D a long way from Q, 
for convenience, but in the formation of the rings the film is 
very thin, and the part of it with which we are concerned is 
very close to Q. 

Just near D we may regard the film as an element with parallel 
faces separated a distance, t, as in the enlarged element. 
MD + DE = 2CD - CM 

2 CN ~, T . . ~ XT cos 2 

= — — - — 2 CN sin = 2 CN . ■ = it cos 0. 
sin sin 

Thus the condition for brightness is : 

2t cos 6 = (2» + 1) -, 


The angle, 0, may be measured by the angle between the 
normal to the plate and the incident rays before striking the 
lens, since near Q the lens acts as a parallel plate of glass very 

It is usual to cause the rays to fall normally on the lens, when 
we have : 

2t = (n + £)x. 

If we consider a plane normal to that of the paper through 
Q and D, we can see that the circumference of a circle with 
centre, Q, and radius, QD, will pass through points at which the 
air film has a uniform thickness, and consequently all round 
this circle rays incident vertically will undergo the same phase 
change, and thus alternate bright and dark rings are formed 
about Q. 

Let the radius of the ring be r n = QD, and let the ring be the 
(» + i)th from the centre. 


The centre itself will be black, for the air film is infinitely 
thin at this point. If this is not the case at first it is because 
some dust particles lie between the surfaces, and these should be 

DQ 2 = DN • DK = t(2R - t), (fig. 193) 

R == radius of lower surface of the lens, 
/. DQ 2 == y„ 2 = 2M, 
approximately, since t 2 may be neglected. 

Thus for brightness : 

r w 2 = R(n + £)x. 

The first ring corresponds to n = ; the second for n = 1, 
and so on. 

Thus the radii are proportional to Vi, V3, V57 etc. 

For the purpose of the experiment a convex spectacle lens of 
about 100 cms. radius of curvature is suitable, and the light 


from a sodium flame is reflected down on to it by means of a 
sheet of plane glass held at 45 ° to the vertical (see fig. 194). 
The rings are viewed by means of a travelling microscope. 

Fig. 194 

In order to focus quickly on the rings, remove the lens and 
focus on the top of the glass plate. On replacing the lens, and 
adjusting the microscope over the point of contact the rings 
which lie in the air film should be distinct. A good bright 
sodium flame is necessary, and often the difficulties disappear 
if this point is attended to. 













■ 1 

(l) denotes the reading on the left of the centre, (r), that on 
the right. 



Move out the micrometer to about the twentieth ring from the 
centre, and then, moving back again, turning the screw always 
one way to avoid any errors due to backlash, set the cross-wire 
carefully down the centre of each bright ring and observe the 
micrometer reading. In this way, by observing the rings on 
both sides of the central dark patch, the various diameters are 
Make a table as shown above. 

Draw a curve with the square of the diameter as ordinate*, 
and the number of the ring as abscissa. The graph should be 
a straight line. 

P 2 N 2 = (D„ 2 )*, 
P X N X = (D n] )*, 
D tt 2 - 4R*(«+i), 
P 2 N 2 - P t N x _ (Pg* - (D Wl )« 

Then if 


N X N S 



If sodium light is used, x may be taken as 5890 x io~ 8 cms. 

In this description we have not taken account of the fact 
that the rings are seen, not directly, but after refraction through 
tlae lens. They are formed in the air film in the space between 
the lens and plate of glass. This difficulty is avoided by placing 
the plate above the lens and in contact with it, for in that case 
we view the rings through a plane sheet of glass. 

Fig. 195 

It is, of course, more convenient to place the glass below the 
lens, for reasons of ease in keeping them steady. The error in 
this case is not great if a thin lens be used, for then the object 
— the rings — is at the surface of the lens and consequently at 
its principal plane. The image is in the second principal plane, 
and of the same size as the object. For a thin lens these planes 
and lens surface are nearly coincident. In practice we have an 


image of magnification slightly differing from unity, or the 
diameters in the formula above are to be multiplied by such a 
fraction. For the purpose of the experiment this factor is 

Jamin's Interferometer 

The apparatus consists essentially of two glass plates, AB and 
A 1 !? 1 , of the same dimensions and optical character. The plates 
are very carefully worked and are of the best optical glass. 
They are mounted parallel to one another, standing on tables 
on an optical bench at a distance apart of about one metre. 

The first, AB, is set at 45 to the bench with its surfaces vertical, 
and it is illuminated by rays from a sodium flame. 





flD S l « 


Fig. 196 

In the diagram a ray, RS, is shown. It is partly reflected 
and refracted at S, and the refracted beam again partly reflected 
at T. The divided ray takes the paths, RSS^HPR 1 and 
RSTUU 1 R 1 , so that it is reunited by the second glass plate. 
This plate is mounted parallel to the first, and set vertically. 
It rests on a table and may be slightly rotated about a vertical 
axis. If both plates are exactly similar and are parallel, the 
lengths of the two paths will be the same for all rays, but by 
slightly rotating one of the plates a difference in path may be 
introduced, differing for different directions so that alternate 
bright and dark bands will be obtained. 

The tube, CD, is then placed in the path of one of the divided 
rays between the plates, the other being allowed to pass clear 
of CD. 

CD is a hollow glass tube with ends of plane, optical glass. 
It is fitted with a tap, E, by means of which it may be exhausted 
or filled with gas. 


The tube is filled with air at the atmospheric pressure, and 
the fringes found by eye. A telescope provided with cross-wires 
is then focussed on the fringes. 

In order to ensure that the fringes arise from interference 
between rays that have passed, one through CD and the other 
outside it, we may cover up the end of CD, and note if they 
disappear as they should if due to this cause. 

In order to be sure that the rays producing interference do 
not both pass down CD, it is necessary to intercept the light at 
the sides of the tube. If they are still present they will arise 
from rays passing down CD. 

When the correct fringes are obtained the cross-wire of the 
telescope is focussed as accurately as possible down the centre 
of one bright band, after exhausting the tube as much as possible 
with an air-pump. The pressure should be brought down to i cm. 
of mercury at least, and a manometer connected up to read 
the pressures. Air is now allowed to pass very slowly into the 
tube by the tap, and the fringes watched. They will appear to 
move across the field of view, and the number passing the cross- 
wire must be counted. Allow about 5 or 6 to pass, and then 
stop the inflow and read the manometer. Repeat this, step by 
step, until the tube is filled with air at atmospheric pressure. 
Draw a graph showing the relation between the pressure within 
CD, and the number of fringes that have passed from the initial 
stage. By an exterpolation deduce the number that pass 
between the limits of complete exhaustion and the attaining of 
atmospheric pressure. 

Let this number be «. 

This means that the difference of optical path in the tube, 
when completely exhausted and when filled with air at 
atmospheric pressure, is wx. 

If the tube is of length, /, and the refractive indices are y. and 
Ho, when the pressure is atmospheric and when the tube is 
exhausted, respectively : 

l(v- — y-o) = ^x. 

But Ho = *» 

so that i* = 1 -\--t-* 

A convenient length for the tube is 30 to 40 cms., when for 
the exhaustion produced by a good air-pump the value of n is 
of the order 200. 

The compensator described in connexion with the experiment 
with Rayleigh's refractometer may also be used with Jamin's 
interferometer, and the calculation of the refractive index at 
normal temperature and pressure may be made with the help 
of Jamin's interferometer also. 


A cylindrical lens with the cylindrical axis vertical is often 
placed just in front of the point S. This has the effect of widen- 
ing the source light in a horizontal direction, leaving it 
unchanged vertically. The source in this case is a slit. 

The Refractive Index of Air by means of the Rayleigh Refraetometer 

A diagrammatic representation of the apparatus is given in 
fig. 197. 

* 0*1 ■ F JEE 3 B aTc~» 

SB p CG 

Fig. 1^7 

ABCD is an airtight metal box, divided into two separate 
chambers, each of which may be connected to a manometer. 
The pressures in the chambers are varied and measured by means 
of the manometers. The chambers are closed at the ends by 
means of parallel plates of good glass. 

The collimator is provided with a slit at K, and provides a 
parallel beam of light which falls on a screen, carrying two fine 
slits, LL, placed in front of the air chambers, so that one slit 
lies adjacent to the end of one, and the other slit adjacent to 
the end of the second chamber. 

These slits are prolonged so as to extend higher than the top 
of the box, ABCD. Thus, light from the slits passes over the 
box as well as through it, and finally enters the telescope, T. 

These two fine slits produce interference fringes in the focal 
plane of the telescope, and if the pressures in the chambers are 
equal, the set of fringes in the lower half of the field appear to 
be continuations of the fringes in the upper half which arise from 
rays that have passed over the top of the box. 

In order to deviate the upper set of fringes down, to make 
comparison with the lower set easy, a prism, P, is provided which 
intercepts the upper set of rays, and deviates them downwards. 

When white light is used, two sets of coloured fringes are 
obtained with a white central fringe. 

If there is a difference of pressure between the two chambers, 
there is a displacement of the lower set of fringes owing to the 
resulting difference of optical path. 

G consists of two plates of glass, inclined at a small angle, 
placed to intercept the lower set of rays. When it lies symmetri- 
cally with respect to the rays striking it, it introduces no 
additional path difference, but on rotating it the rays through 
one plate traverse a longer path within the glass than do those 
in the other. 


In this way the lower central band can be moved about, and 
we may, by a rotation of G, bring back the fringe to its central 
position after it has been displaced on account of the difference 
in air density in the two chambers. G carries a pointer moving 
over a scale, and we may calibrate the scale so that the difference 
of path introduced by setting G is known. Thus, by altering 
the pressure within the chamber and then moving G to counteract 
the displacement of the central fringe, we can read directly from 
the calibrated scale to how many wave lengths the path difference 
in the two chambers amounts, and a graph is plotted with the 
pressure differences as ordinates and the number of wave lengths 
as abscissae. 

The calibration is performed by making the pressure in both 
parts of the chamber equal and illuminating K with mono- 
chromatic light, e.g. by a sodium flame. 

The scale is set so that the bright bands of the upper set of 
fringes lie over those of the lower. The pointer is then moved 
over the scale until a certain number of bands pass a fixed point 
in the upper series. The following table will illustrate how this 
result should be recorded : 





















4'8 9 






























Mean for 40 



wave lengths 
Mean value of 


1 wave length 

in scale divi- 



After the calibration of the scale the collimator slit is again 
illuminated by white light, and the central fringes arranged one 
above the other. 


The apparatus is shorter since the length of liquid traversed 
need not be so great as that required for a gas. 

If a liquid is placed in one of the vessels, and the same liquid 
containing a solvent in the other, we may determine the effect 
of the solution on y., or inversely, we may estimate the amount 
of solvent contained from its refractive index. 

Fig. 202 

In this instrument the movable plate is in the upper beam 
•instead of the lower, but the method of use is otherwise similar 
to that of the previous experiment. 

Two thermometers project into the liquid for recording their 

Examine the changes produced by adding small quantities of 
a salt to pure water, and placing in one cell the solution so 
obtained and water in the other. 

We have, if y. and tx 1 denote the refractive indices of water 
and of the solution respectively, d the thickness traversed by 
the light, and x the wave length : 

where n is the number of displacements of the central fringe. 
The experiment is performed with white light so as to have a 
definite central fringe, but x is the wave length of the light used 
to calibrate the scale. By moving the pointer over the scale 
until the two central fringes are coincident, the scale reading 


the tubes containing the gas, which would normally separate 
the lower interference bands from the comparison upper bands, 
produce no image in the field of view on account of the refraction 
in H. The two sets of fringes stand one immediately above the 

They are focussed by the achromatic lens, Q, and examined 
by the cylindrical lens, R. This provides a large horizontal 
magnification of the fringes, which lie close together because 

Fig. 201 

the slits are of necessity rather wide apart (cf. formula 6 of 
this chapter, d is large in the present case). The lens, however, 
does not give magnification in a vertical direction, so that the 
shadow cast by the obstacle is not broadened vertically, and 
there appears a sharp dividing line between the two sets of bands. 

The Rayleigh Refractometer for Liquids 

Fig. 202 shows a Hilger apparatus based on the foregoing 
principles for comparing the refractive indices of liquids. 


If T is the absolute temperature, we have also : 

—, = constant. 

Thus — - — . T is constant. 

If jx is the refractive index of air at N.T.P. 

{* — i . T _ t*o_— i 

x 273. 

P 76 

We shall see that y. can be determined by observations in this 

Suppose the length of the tubes containing the air to be I, 
and the refractive indices, jx x and [t it corresponding to pressures. 
p x and p 2 . Let the wave lengths of monochromatic light be X, 
and X, respectively. 

The excess of waves in one tube over those in the other is : 

' (t t) ". G£ ~ ■£) = i ( "' - " i) 

This number is deduced from the scale over which the pointer 
moves. Suppose it is m. 
Then : 

T ■ 76Tx 

273l(p2 - Pi) 
The pressures are measured in centimetres of mercury, and the 

ratio, -r — , is deduced from the graph as described, while the 

length of the tubes, which is usually about 25 cms., may be 

measured by means of a metre rule 

B C 

D * E 

Fig. 199 

In figs. 199 and 200 a modern form of apparatus is shown 
diagrammatically. The plates, L,K, correspond to G, and the 
prism, P, is here denoted by H, and its action shown in the 
fig. 200. It can be seen how such obstacles as the upper edge of 


The pressures in the chambers are varied and measured by 
the manometers. 

The central fringes are kept one above the other by means 
of G, and the pressure differences recorded along with the position 
of the pointer on the scale. 

Draw a graph with scale readings as ordinates and pressure 
differences as abscissae (fig. 198). 

O 5 lO t5 ZO 

Pressure Difference. 

Fig. 198 

White light is used in order to provide a definite central fringe. 
The position of this fringe is independent of colour while all 
other fringes have positions which depend on the wave length. 
If monochromatic light is used, all the fringes are alike, and the 
central fringe is indistinguishable from the others. 

Since the screw attached to the compensating glasses has been 
calibrated in wave lengths of sodium light, the observations give 
the path differences in terms of so many wave lengths of yellow 
light, and the refractive index deduced will be that for this 
particular wave length. 

Fig. 201 gives a general view of the apparatus. 

From the slope of this graph may be deduced the difference 
in path in wave-lengths for a difference of pressure of 1 cm. of 

Observe the temperature of the air in the tubes by placing a 
thermometer close to them and noting its indication throughout 
the course of the experiment. When the pressure is varied by 
means of the manometers, the changes should take place slowly, 
and time should be allowed for the air to take up atmospheric 

Theory of the Experiment 

For a gas the relation, 

u — 1 

= a constant, 

is very approximately true. We assume it in this experiment, 
(i is the refractive index, and p its density. 


will give the number of wave lengths which one ray has fallen 
behind the other in traversing a path of different nature from 
that of the other. 

The calibration is carried out in a preliminary experiment 
with both liquid cells containing water and with a sodium flame 
illuminations, as in the last experiment. 

Michelson's Interferometer 

The apparatus is illustrated diagrammaticaUy in fig. 203. 
It consists of two plane mirrors, M x and M 2 , silvered on their 
front surfaces and mounted vertically on a heavy, firm, rigid 
stand. The stand consists of a metal bed provided with a large 
micrometer screw of very fine pitch. Rotation of the screws 
causes M t to slide along the bed and its position may be read 
off at the screw-head. A general view of the apparatus is given 
in fig. 204. 



The second mirror, M 2 , is fixed at the end of a metal arm 
mounted at the end of the bed, and at right angles to it. This 
arm also carries the two sheets of plane optical glass, P a and P 2 , 
which are equally thick, and are mounted at an angle of 45 to 
the arm. 


In order to produce interference fringes, a source of light, e.g. 
a sodium flame, is placed at the focus of a lens, L, and a parallel 
beam of light thrown on to the glass, P x , the direction of the 
beam being along the arm carrying M 2 . One of the rays of such 
a beam is shown in the figure as SN. On striking P x it undergoes 
partial reflection and refraction at N, and the refracted part is 
again divided at K, aDd later further division takes place at P 2 . 

Fig. 204 

We consider one of the ways in which division can take place, 
the recombination of two parts, which have started from one 
ray and have different optical paths, producing interference. 

A marked difference in this case from that in some of the 
foregoing is the production of interference between rays differing 
in phase by very many wave lengths. In the case of Newton's 
rings or Fresnel's biprism, the path difference amounts to a few 
waves only. 

The ray, SN, is refracted in P x and reaches K, where it is partly 
reflected so that it gives rise to KQT and partly refracted to 
give KM 2 . These rays are both returned along their paths and 
reunite to produce KW. In the course of their journey each 
passes through the glass a distance equal to 3NK, and change 
in phase is brought about by the difference in the path in air. 
This may be varied by altering the position of M x . 

The ray, KW, may be observed by eye, and usually curved 



interference bands similar to those observed in the experiment 
on Newton's rings will appear. The curves will not be seen 
closed, only parts of the closed curves can be seen. 

We use here, as in the experiment on Newton's rings, an 
extended source of light, for the rays from each point of the 
source are divided and re-combination takes place between rays 
which originally belonged to the same ray. Rays falling on P x 
in the same direction all suffer the same phase change and emerge 
parallel to KW, so that an optical instrument will focus them 
together in its focal plane. 

Other rays in a direction slightly inclined to SN, emerge as 
a set of parallel rays, slightly inclined to KW. These undergo 
a different phase change, and also lie in the focal plane displaced 
from the image due to rays parallel to KW. 

It is necessary that the mirrors, M x and M 2 , should both be 
vertical, and in order to allow this adjustment to be made, at 
the back of M 2 are three screws pressing against springs that 
cause it to rotate. 

If a sheet of tin carrying a fine hole is placed in the path of 
the incident light, four images are seen as a rule, when the eye 
looks in the direction, WK. The reason for this is best seen by 
reference to a diagram (fig. 205). 

The emergent rays are marked p lt p it p s and p lt and it will 
be noted that p 2 passes through the glass plates three times; 
so also does p x . 

Thus if the rays are made to coincide they will be in a condition 
to annul or reinforce each other according as the path difference 
is an even or odd number of half wave-lengths. 

The rays, p a and p x , traverse the plates five times and once 


If a card is placed in front of M x , the rays p x and p s still 
appear ; and p x will be the brighter since it traverses the plates 
only three times, so that it can be distinguished. 

By placing a card in front of M 2 , the rays are cut out except 
p a and pi, so that by this means the images can be distinguished, 
and those due to p x and p 2 caused to overlap by adjusting the 
screws behind M. x and M 2 . 

When both mirrors are vertical the images coincide in pairs. 
One constituent of each double image lies behind the other. 
Those which produce interference lie along the direction, WK. 


Fig. 206 

Fig. 206 illustrates the mode of production of the fringes from 
another point of view. The interferometer acts as if we had a 
source, S, from which rays could be reflected by two parallel 
mirrors, M x and M 2 , in which they would produce images, I x 
and I 2 . The distance, MjM 2 , is equal to the difference of the 
distances, NMj and NM 2 , of fig. 203. 

At any point, P, rays, SAP and SBP, would unite, and if their 
paths differed in length by wx, a bright point would arise. 

This is equivalent to stating that in this case : 

PI 2 - PIj = »x. 

Thus we obtain the particular interference band at all points, 
P, for which this relation holds, i.e. P lies on a hyperboloid of 
revolution with SM X M 2 or with WK, of fig. 203 as axis. In a 
plane perpendicular to that of the figure we have a circular 
section of the hyperboloid which explains why the fringes are 
circular as seen on looking along WK. 

The Determination of the Frequency of Light from a Sodium 
Flame or any Monochromatic Source 

A striking feature pi the Michelson Interferometer is the screw 
which displaces M v In the apparatus illustrated in fig. 204 the 
screw has a length of 200 mm. and a pitch of 1 mm. The head 
of the screw is furniihed with a scale divided into one hundred 
parts, each thus corresponding to one hundredth of a millimetre. 
The screw may be rotated by the handle seen in front of the 
apparatus. The sm^ll lever on the right of the front of the 
apparatus puts into action the slow motion screw, one turn of 


which corresponds to one division on the head of the main screw. 
As the head of the slow motion screw is also divided into one 
hundred parts, it is possible to record a motion of the mirror of 
one ten-thousandth of a millimetre. 

In order to obtain the fringes, set up the sodium flame at S, 
and place immediately in front of it a sheet of tin with a small 
hole in it just opposite the bright part of the flame, and adjust 
the flame and hole to lie on the level of the centre of the mirrors 
and plates. 

Mount the lens, L, also so that its centre is at the same height 
as the hole, and place a piece of plane mirror between L and P a 
and move L until an image of the hole is thrown back on to the 
tin close to the hole. The light is then parallel as it leaves L. 
Remove the plane mirror and look in the direction, KW, when 
usually four images of the hole appear. Adjust the mirrors so 
that these images coincide, two by two, as indicated above. 

Then remove the tin sheet ; allow the light from the flame 
to fall through L on to the apparatus. 

As a rule very slight movements of the mirrors will bring the 
fringes into view if they are not already to be seen. Sometimes 
a slight motion of M x along the bed of the apparatus helps to 
discover the fringes. 

Set up in front of P 2 on the side towards W a sharp pin point 
to mark the position of the centre of one fringe. Rotate the 
screw slowly by the slow motion screw, and watch the movement 
of the fringes across the field of view, counting the number which 
seem to pass the point, and observing from the scales how far 
M x has moved. 

When Mj moves back a distance £x the path difference between 
the two rays which unite to interfere has been increased by x, 
so that where a particular fringe originally appeared the neigh- 
bouring fringe now apparently lies. 

Thus if Mj moves a distance, I, the number of fringes which 

appear to move past the point is correspondingly — . If these 

are couhted, since / can be measured, we can find x. 

Do this for such monochromatic extended sources as are 

Note also that the intensity of the fringes appears to alternate 
as the distance of M x varies. We shall make use of this fact in 
the next experiment. 

The Determination of the Difference of Wave Length for the 
Sodium D Lines. 

Adjust the mirrors Mj and M 2 so that their distances from N 
are equal, as nearly as can be judged by eye. 


This adjustment may be brought about accurately by observing 
the images of the holes formed as above in the two mirrors, and 
adjusting M x until there is no parallax between them. In this 
case they are equally distant from the observer, and the two 
mirrors are consequently equidistant from N. 

The light from a sodium flame, though for many purposes 
considered monochromatic, contains two fairly intense waves 
whose frequencies are close together. 

In the present case both sets of fringes which arise from the 
two waves overlap, but if M x is slowly moved away there is a 
gradual separation of the two sets, and finally the bright band 
of the one lies over the dark band of the other. This happens 
when the distance that M x has moved from the first position 
contains one more quarter of a wave of the one than of the 
other, for then a difference of phase, corresponding to one-half 
wave length has been added to one more than to the other. 
Or, to put it otherwise, let I denote the distance moved by M lf 
the additional air path added to each wave incident on M x is 
thus 2.1. 

Suppose 2.1 contains n x waves of length x x , and n 2 of lengths x a . 
Then the difference between n x and n t is one-half. 

Suppose for the sake of definiteness that n x > n t and conse- 
quently x x < x t . 

Then, since n x = —and », = — , 
Xj x 8 

we have : 2l( J = -, 

\X X Xj/ 2' 

i i_ i . 

Xj x a 4' 

If the two waves had equal intensity the field would become 
uniformly illuminated, and the fringes would disappear. 

In this case, since one of the lines is more intense than the 
other, we get an alternation in distinctness, the brighter fringes 
still stand out in contrast with the adjacent less bright ones. 

Note the positions of Mj at the beginning and successively at 
positions where the fringes become least distinct and again 
distinct as M x goes further away and the path difference contains 
one complete wave more of one colour than of the other. Do 
this for as many cases as possible, and if d denote the distance 
between the positions of M x in which two successive distinct 
sets of fringes occur, we have 

j_ i_ __ i 

Xj X a 2d' 

Assume the shorter light in the sodium light to have a length 


5890 x 10- 8 cm., and deduce the difference between x x and x 8 
in this case. 

The Production of Coloured Fringes 

These can be obtained when the mirrors are set for equal 
paths. Do this as in the last experiment with sodium light and 
then replace it by a white source. The fringes due to the different 
colours overlap in this position. 

With sodium light, as M x is moved, it will be noticed that the 
circles change the direction of their curvature ; the position 
required is just when the transition takes place. It is difficult 
to decide just when this occurs, and it is a good plan to note 
the two positions of M t in which the fringes are definitely curved 
in one way and then in the other. Put in white light when M x 
occupies the position corresponding to one of these directions 
of curvature, and then slowly move it back to the other position. 
The coloured fringes will appear during this movement. 

M, must be moved slowly or the fringes change their places 
too rapidly to be noted. 

Fig. 207 

The Hilger Wave Length Spectrometer 

This apparatus, illustrated in fig. 207, consists of a cast-iron 
stand with two arms at right angles on which are held rigidly 
the telescope and collimator. 


The telescope may be fitted with a high-power eyepiece with 
adjustable cross-webs or with a shutter eyepiece which can be 
adjusted to cut out any part of the field except that under 
particular examination with an adjustable metal pointer. The 
pointer has a brightly polished fine point and is illuminated by 
reflecting light from outside by means of the mirror shown in 
the figure in position above the eyepiece. Thus any point in the 
field may be taken as a reference point by setting the pointer 
to it (fig. 208). 

Fig. 208 

In addition the shutter eyepiece may be employed with light 
filters which impart any desired colour to the bright point. 
This adds to comfort in reading and consequently to accuracy. 


Fig. 209 

The vertical collimator slit may also be reduced in length by 
means of a cross horizontal slit, so that a small rectangular 
source is obtained. 


The principle of the apparatus is based on the constant devia- 
tion prism which is illustrated in fig. 209. The faces particularly 
concerned in the deviation of the ray are inclined at the angles 
marked in the figure, and total reflection occurs at the face AC. 

The prism is mounted on a turntable in a position marked 
for it. The mean deviation of the rays is a right angle, and in 
order to pass through the spectrum the table is turned by means 
of a screw to which a drum is attached provided with a milled 
head (fig. 210). On the drum is a scale so that the wave length 
of any line under observation and appearing in the field of view 
of the telescope may be read off directly. 

Fig. 210 

Before taking any observations of wave length it is necessary 
to adjust the prism accurately so that the correct wavelength 
is indicated when the corresponding line appears at the eyepiece 

To make this adjustment, illuminate the slit with light of a 
standard wave length, set the drum so that the appropriate 
wave length is indicated at the index of the drum and adjust the 
prism so that the line appears under the eyepiece index. 

Clamp the prism in position with the screw provided. 

Other wave lengths of light illuminating the slit may then be 
determined by rotating the prism by means of the drum until 
the line appears at the eyepiece index and reading off the number 
against the drum index. 

Fig. 209 indicates the course of a ray in the prism. 

In the apparatus designed for use with certain accessories — 
the Lummer-Gehrcke Parallel Plate, the Fabry-Perot Etalon or 
the Michelson Echelon Grating, the collimator arm is of greater 
length than that illustrated in the figure to permit of interposing 
the accessory between the prism and collimator. 


The method of fixing the collimator and telescope and of 
obtaining different parts of the spectrum by rotating the prism 
is very convenient and accurate ; moreover, the drum can be 
rotated while looking through the eyepiece, and one is saved the 
inconvenience of moving round with a rotating telescope. 

A suitable standardizing wave length is the red line of the 
helium spectrum, which has a wavelength of frequency 6678-1 
Angstrom units (1 Ang. unit = io -10 metre). It is the shorter 
of the two red helium waves. Or the sodium lines may be used. 
These are separated by the prism and have wave lengths 5890*2 
and 5896*2 Angstrom units respectively. It is a good plan to set 
the instrument on one of these lines and check the setting by 
turning to the other and noting if the reading gives the correct 
wave length. 

The slit may be illuminated by throwing an image of the 
source on it. In the case of the sodium lines the source is 
obtained in the us,ual way, and for a helium line throw the image 
of the bright part of a helium discharge tube on the slit. 

A protective metal cover for the prism table is provided. 

When a photograph of the spectra is required the eyepiece is 
removed and replaced by a camera with a suitable focus lens. 
This is shown in fig. 211. It is capable of adjustment by tilting 
so that the whole spectrum can be photographed and a vertical 

Fig. 211 

displacement enables the same plate to be used giving photo- 
graphs one above the other. There is, of course, also a shutter 
for exposure. 

In addition to the usual vertical slit there is a horizontal 
adjustable slit attached to a hinge so that it can be swung out 
of the way or in a position covering the vertical one. 


Thus we have crossed slits, and a small rectangular source can 
be obtained by closing down both the slits as much as is required. 

The Lummer-Gehrcke Parallel Plate 

This piece of apparatus 1 s described in the " Annalen der 
Physik," vol. 10, 1903, p. 457, and a very complete account of 
its use and theory given. 

The reader is recommended to refer to the original paper, 

but we shall give below as much as is necessary for our purpose. 

The plate can be used in connexion with the Hilger Constant 

Deviation spectrometer, and is of quartz of refractive index 1*544. 

It has the following dimensions : 

Length 130 mm. , 
Width 15 mm., 

Thickness \\ mm. (approximately). 

Like the echelon grating it produces spectra of a high order, 
and has consequently a high resolving power. The approximate 
resolving power in the present case is about 200000. 

It is therefore well adapted for the study of complex lines 
in the spectrum which in lower orders appear as single lines. 

It may be used also to measure such small displacements as 
occur under the influence of magnetic fields, and in the section 
immediately following the present description we describe how 

it may be used to determine the value, — , from observations on 
J m 

the Zeeman effect. 





Fig. 212 

The plate itself is shown in fig. 212 and mounted on its stand 
in position on the spectrometer in fig. 213. In this figure the 
screws permitting adjustment in the various directions are shown. 

The action of the plate is illustrated in fig. 214. Here LM X 
denotes an incident ray of monochromatic light making an angle, 
i, with the normal to the plate. After refraction the angle is r. 

The figure shows the production of two beams emerging from 
the plate on opposite sides. The rays, M x i, M 2 2, M33, etc., and 
Nil 1 , N 2 2 x , Njtf 1 , etc., on account of their different courses due 
to successive reflections and refractions, are in different phases 
when they reach the position denoted by 123 and zH 1 ^ 1 respec- 
tively. These two traces mark out wave fronts and the emergent 
beams produce interference bands. The upper beam consists of 
a system of bands with alternating intensities, the maxima 


having an intensity which may be measured by 2 J, while the 
minima have intensity, J. Thus the effect is the same as is 
obtained by imparting to the whole field an intensity, J, and 
drawing bands across it of double the intensity. The transmitted 
system, however, consists of a set of maxima of intensity, J, 
with alternating minima of nearly zero intensity. For con- 
venience in observation the second system is to be preferred, 
and, by using light at grazing incidence so that i is a right angle 
and r the critical angle for quartz the sharpness of the lines is 

Fig. 214 

It is not possible to give here the reasoning which leads to 
these statements ; the discussion is given very clearly in the 
original paper. 

These points find application in the Hilger pattern. 

A slot will be noticed on the right of the carrier of the plate, 
fig. 213. Just opposite this is a prism of such dimensions that 
the light along the directions MiNj, M a N 2 , etc., is in the critical 

The prism lies on the under surface of the plate, so that the 
beam emerging from above is the transmitted beam. 

The other beam is absorbed by the black lining of the stand 
on which the lower surface of the plate rests. 

In fig. 212 the prism is represented at OQ. 

The Theory of the Plate 

Corresponding to every direction, *", of incidence there will 
be in the wave front, i^^ 1 . . . , a particular variation of 
phase from point to point on account of the different courses 

I'UV :,.:i. 


taken by the rays. These rays are received by some optical 
instrument, for example, the telescope of the spectrometer, and 
focussed in the focal plane. Corresponding to the waves drawn, 
we shall have a point image in the focal plane. These rays, 
however, are those lying in the plane represented by the paper. 
Above and below them lie rays coming from rays above and 
below LMp and parallel to LM X , whose courses are exactly 
similar, so that in the focal plane above and below this point 
image lies a series of points forming a line. We shall work out 
the intensity corresponding to this line. If we consider a slightly 
different direction of incidence, i + 8i, the emergent beam is 
also slightly different in direction, and in its course through the 
plate. The phase differences in the corresponding wave front 
will thus be different, and the corresponding line in the focal 
plane will have a different intensity. 

A wave travelling along a direction denoted by r will at a time 
t, be represented by : 

a sm 2tc 

(t -{) 

Here T is the period of vibration in the wave, x the wave-length, 
and a the amplitude, t is the time measured from some con- 
venient instant, and r the distance from a convenient origin. 

In our case the origin will be conveniently the point, N x , 
and we shall consider the wave motion corresponding to the 
position, i^^ 1 , at which the time is measured by t. 

On refraction into the plate there is a diminution of intensity, 
and since the intensity is proportional to the square of the 
amplitude of the wave motion we can express this by regarding 
the refraction as causing a diminution of amplitude, so that 
amplitude a in the air becomes sa in the quartz, where s is a 
fraction. On passing out at Nj, the amplitude is again reduced 
by a fraction, s 1 . The amplitude for the ray Nji 1 is thus 

We may not suppose that s and s 1 are the same, since in one 
case the refraction is from air to quartz, and in the other from 
quartz to air. 

There is also a change of intensity on reflection at the points 
M 2 , M 3 , M 4 , etc., and N x , N 2 , N 3 , etc. We shall suppose this to 
diminish the amplitude each time by a fraction, a. 

Thus the ray, N^ 1 , has undergone refraction at M x and N 2 , 
and reflection at Nj and M 2 . The incident amplitude, a, is thus 
reduced to oHsfy ; and similarly the amplitude of N33 1 is 

The disturbance which gives rise to the ray, N 1 i 1 , starts out 


from N x with an amplitude, ss^, and at i 1 , since it has travelled 
a distance, r lt we may represent the displacement by : 

ss^ sin 2tz 


Denote the foot of the perpendicular from N 2 on N^i 1 by A, 
and from N 3 on N 2 2 x by B, and denote the equal distances, 
N 2 A, N 2 B, by e. Let the distance, MjNj be denoted by <5, 
and the refractive index of the quartz by [*. Then the distance 
traversed in quartz is equivalent to the distance, [l6 in air. 

Thus the point 2 1 is at an equivalent distance [r 2 + 2jx<5) 
from N lf and since r t = r 1 — e we may denote the displacement 
at 2 1 by : 

„ , . ft r x — c + 2fi<5\ 

aHsH sm 27tf ^ * — J ; 

and similarly the displacement at 3 1 is : 

. . ft Y x — 2e + 4tx<5\ 

o^s 1 ^ sin 2rc( — 5 x ) 

There is, of course, an indefinitely large number of such terms 
as the three given above, and since all these rays are focussed 
at a point in the focal plane of the telescope, the total effect at 
this point is obtained by adding together all the terms. 

In practice we shall have a small parallel bundle of rays 
falling at M lf giving rise to small parallel bundles at N x , N 2 , N 8 , 
etc. Throughout the bundles there is, however, the same phase, 
as may be seen by considering a ray parallel to LM X , falling at 
any angle, i, to the normal. Thus the total effect is merely 
multiplied by some constant factor on account of the incidence 
of more light than that we have supposed is represented by the 
ray LM X . 

Thus we have a total displacement, Y\ where : 

Y = SS %sin 2,(| -£)■+ «Wsin a, (I- r - ~ ;+ 2 »> ) 

, . . ft r x — 2e + 4tid\ 
-|- ss x a<s* sin 2rc(=r - 1 + ... 

A typical term may be written : 

»>..» sin a, j (i - &) + p ■ i^| , 

where p has the values, o to infinity. * 

If we write a = 2ir (f ~ ^) ' P = T ^ ~~ ^' 

we have : 

Y = Xss l a{o& sin (a - p$)} (7) 

* In practice the upper limit to the value of P is about 15, the dimensions of 
the plate permitting about 15 reflections. The terms beyond the fifteenth are 
email and contribute but little to the value of Y. 


The sum of this series is finite since a is a fraction, and has 
the value 

sina — o 2 sin (a + ft) 
1 — 2a 2 cos ft + o 4 
For the proof of the summation the reader may be referred to 
Hobson's " Plane Trigonometry," §76, p. 91. 
Thus the resultant displacement is : 

_ r , sin a — a 2 sin (a + ft) 

1 — 2a 2 cos ft + o 4 

It will be noted that for any particular direction through the 

plate the quantities, e and 6, are constant, so that ft is a constant. 

s, s 1 and a are also constant terms so that it is only a that 

contains the variable time, t. 

sin a — a* sin (a + ft) = sin a (1 — o 2 cos ft) — cos a • a 2 sin ft. 

We may write this 

= A sin (a. — <f>) 

■u a. c 2 sin ft 

where tan <t> = _— , 

1 — o 2 cos ft^ 

and A — Vi — 2a 2 cos ft 4- 0*/ 

===== ' sin (a - 

2a 2 COS ft + o 4 

Thus Y = . • sin (a — tf>) 

Vi —2a 2 COS ft + o 4 

sra lT-* — r) 

Vi — 2a 2 COS ft + o 4 

This denotes a simple harmonic vibration of amplitude, 

55 x a 

Vi — 2<J 2 cos ft + a 4 ' 

and it measures the amplitude on the line in the field of view 
of the telescope due to the reception of the emergent beam 
along the direction which makes an angle, i, with the normal 
to the plate. The intensity is proportional to the square of this 
amplitude, and is therefore a maximum when (1 — 20 2 cos ft+o 4 ) 
is a minimum, and a minimum when this expression is a maximum. 
But 1 — 2« 2 cos ft + c 4 = (1 — a 2 ) 2 4- 4a 2 sin 2 Jft. 

This has a minimum value when sin \ ft has the value zero, 
and a maximum when sin \ ft has the value unity. 
Thus lines of maximum intensity correspond to the value : 
ft = o, 2k, 47c, etc., 
and lines of minimum intensity correspond to : 

ft = *> 3*> 5"» etc. 

But ft = (2(i<J — e). 


Thus for a maximum, 

2[x<5 — e = n\, 
and for a minimum, 

2y.8 — s = (n + £)x, 
and n may have any integral value. 

If d denote the thickness of the plate : 
8 — d sec r, 
and c = MiMa sin * 

= 2d tan r sin i = 2^ tan r • {/. sin r. 

Thus 2{A<5 — e = 2yd cos r, 

and the bright bands lie in directions given by : 

2yd cos r = n\. 

By proceeding further with this discussion, Lummer and 
Gehrcke have drawn the conclusions, to which we have referred 
above, concerning the intensity of the bands. 

Students who prefer it may substitute the following proof 
which enables the same formula to be derived shortly, but in a 
way that does not lead to any expression for the intensity of the 

The path difference between any two of the rays, say Nil 1 
and N 8 2 x is equal to : 

2(X(5 — e = 2yd cos r. 

We may take all the rays emerging from the plate in pairs 
which are separated by. the same distance, equal to N X N 2 . 

These therefore interfere and will produce darkness or bright- 
ness, according as this phase difference is equal to wx or (w+|)x, 
where n is a whole number as before. 

Thus for bright lines : 

2yd cos r = n\. 

If we refer to fig. 213 and note the position of the plate it is 
clear that the different orders come out from the plate above 
one another. As the path difference increases the angle of 
emergence decreases, so that the higher orders will lie higher in 
the field of view than the lower in the case of a telescope with 
an erecting eyepiece. 

In our case the mean angle, r, is the critical angle and there is 
grazing incidence. The different orders correspond to angles 
very slightly differing from grazing incidence. 

By substituting the value, y. = 1*544, and the value for the 
critical angle, r = 40°22 ', together with the value of d given above, 
we find that for X = 5890 tenth metres (10- 10 metre), the order 
is approximately 18000. 

In order to set the instrument in position, the spectrometer 
is first set up in the manner previously described, and the slit 
illuminated by means of some convenient monochromatic light. 

Paga 351 


The plate is placed in position and adjusted by the screws until 
the brightest image is obtained in the eyepiece. Since the 
orders lie one above another, a vertical slit cannot be used, for 
the different orders will appear overlapping the image of the 
slit. Thus the crossed slits must be used with a small rectangular 
source. The length of the slit need not be very small, for on 
account of the dimensions of NO, fig. 212, only a fraction of the slit 
is effective in producing bands. The slit should be small enough 
to avoid overlapping, but wide enough to produce intense bands. 
The images may then be viewed or photographed as desired. 
We have throughout disregarded the possibility of a change of 
phase on reflection at the two surfaces. 

The Fabry-Perot Etalon. (" Annates de Chimie et de Physique" 
1897, Ser. 7, p. 459.) 

This apparatus is another means of obtaining a high resolving 
power, and is in many respects similar to the Lummer-Gehrcke 
Plate. Fig. 215 shows the apparatus in position on the arm of 
a constant deviation spectrometer. 

Fig. 216 shows the two plates, ABCD and FGHE, of which 
it is composed. These are placed accurately parallel, and are 
separated by a distance piece consisting of a hollow cylinder of 
fused silica. This substance has an extremely small coefficient 
of expansion, so that the distance between the plates may be 
regarded as independent of the temperature. 

A B E r 

D C H G 
Fig. 216 

The faces, BC and EH, are silvered by cathodic deposition 
in order to increase their reflecting power but so as to leave 
them partly transparent. 

Light entering the plates therefore undergoes multiple reflec- 
tions between the silvered faces and produces also partial 
transmission through the opposite face. 

The faces, AD and FG, are inclined to CB and EH in order 
to avoid interference effects that would occur through multiple 
reflections and refractions if all were parallel, but AD and FG 



are parallel, so that light incident on one side leaves the other 

Fig. 217 illustrates the action of the plate. The air space 
between the plates is shown, and the course of a ray as it is 
multiply reflected between the plates. The refracted portion 
leaving EH is shown, but the refraction at the other face is 
omitted for convenience. We consider an incident ray, LM X , 
from the instant of its arrival at a point, M v within the first 
wedge and immediately before refraction into the air space. 


^ Y 
Fig. 217 

The wave emerging within the second plate is shown, and 
XY denotes a wave front. 

The wave is refracted out at the surface, FG, into the air, 
but as no further change of phase occurs after the light leaves 
the air film, we do not need to consider the wave beyond XYZ. 

Let the wave on starting out from the point, M x , have, in the 
air space, an amplitude, a, and let its period and wave length 
be T and x, respectively. Whenever the wave passes from the 
air space into the second plate, let the amplitude be reduced in 
the ratio, 0, and on reflection at the partially silvered faces let the 
amplitude be reduced in the ratio, /. Both these quantities are 
positive proper fractions. 

Thus the wave proceeding along M X N X may be denoted by : 

y = a sin 2n ( - — -J, 

where r measures the distance from the point, M x , at which the 
displacement, v, is considered. 

Let the equivalent path difference between M x and the points, 
X, Y, and Z, differ successively by an amount, 6. 


Then proceeding as in the theory of Lummer-Gehrcke Plate, 
we have : 

8 = 2e cos i, 

where e is the distance between the plates, and * the angle 
of incidence at the surface, EH. 
Thus the displacement at X may be written : 

da sin 2n 

\T x> 

where r x denotes the equivalent air path between M x and X. 
The displacement at Y, which after two reflections has been 
reduced by the ratio, / 2 , and for which the equivalent path is 
('1 + <5), is given by : 

and similarly at Z we have : 

tf% ana. (i - i±»?). 

Of course there are many more points such as X, Y, Z ; and 
if we continue the series of terms until the contributions become 
negligible, and if we view the rays in a telescope, we have, for 
the total displacement in the field of view of the telescope, a 
quantity : 

Y = da |sin27c (y~^) +/ 2 sin27r (i-^ 1 - a) 

+/ 4 sin 2k (| — £ - 2AJ + . . .J, (8) 

where A is written for - •■ 

Strictly, A ought to include a quantity due to change of phase 
on reflection at the air-glass surfaces ; if this is w then A — - + w. 

In order to compare this with equation (7) above, we may 
note that the quantity, a, is the same for each, that 6a takes 
the place of ss^, and that / takes the place of o, while 2* A 
corresponds to p. 

If, therefore, we write p = 2tcA, equation (8) may be written : 

Y = S Oaf* sin (a - p$) (9) 

where p has the values, o to infinity. * 

Thus as before, the value of Y is : 

, sin j 2ir ( — — -^ ] — </> I (10) 

Vi -2/ 2 cos p+/* ( \T x/ / K ' 

2 3 *Strictly p has an upper limit between 10 and 20. See p. 348 


where : 

. . /a sin , x 

tanft = J ~ a .. . (ii) 

i — / 2 cos p x ' 

Thus the intensity of the line appearing in the focal plane of 
the telescope, corresponding to this particular direction, is propor- 
tional to the square of the amplitude of this expression, or we 
may measure the energy by the expression : 

(i -2/ 2 cosp+/*)' 

This may be written : 

2 fl a 

(i -f 2 )'' 

1 + ( T^p sin2TCA 

Thus the intensity fluctuates between a maximum and mini- 
mum value, the former of magnitude : 

fl 2 a 2 
(i -/ 2 ) 2 " 

This may be large provided that the value of / is not far from 

Denote this by I . 

Then we have for the intensity : 


I = 

I + (Wr sin27cA 

The minimum of I corresponds to the value for A which makes 

sin 2 reA = i, 
and the magnitude is consequently : 

(h^ < a > 

Now with / not very different from unity, this may be a very 
small fraction of I , so that there is a great contrast between 
the maxima and the minima. 

Another point may most conveniently be brought out by a 
numerical example. 

Suppose / = '87 = ( y^ ) — this value has been chosen 

for convenience. 

1 = h 

1 -f 48 sin 2 7cA 


A maximum occurs when A = m, and the next when 
A = (m + 1), tn denoting an integer. 

Consider the case when A = m -f ^, 

i.e. when we have gone over ■- of the interval between the two 
values of A. 
In that case it follows that : 

I = 


1+48 sin 2 — 

= i I , approximately. 

This means that the intensity falls off quickly as the maximum 
position is left, so that we have bright lines in the field of view 
separated by a comparatively long dark interval. 

Thus, if there is a second ray in the field of slightly ^different 
wave-length, its lines will not overlap those of the other ray 
unless there is very little difference indeed between the two 

From the above theory we see the influence of the partial 
silvering in producing sharp bright lines on a background that 
is almost black. If a layer of air is used between two plates 
without any silvering, the reflecting power is small, and we have 
fringes produced and superposed on a field of uniform illumination 
with no comparative broad spaces between the fringes. 

We can explain these points quantitatively by the aid of 
some numerical examples given in the original paper. 

It is usual to speak of the reflecting power of a surface and 
not of the quantity we have denoted by/. But since the intensity 
of a ray is proportional to the square of the amplitude, and since 
the amplitude on reflection is reduced by /, we have, if the 
reflecting power is denoted by R, 

so that the expression denoted by (A) above gives for the ratio 
of the minimum to the maximum intensity : 


If R = -042 

P = -84, 

so that for a small reflecting power the maxima and minima 
have nearly equal intensity. 

If, however, R = -74, 

P = -02, 

and the minima are very feeble. 


The way the intensity falls off rapidly as the maximum is 
left has already been shown. The value for / taken above, 
corresponds to a value of R = 75, approximately. 

Fig. 218 shows graphically the difference in the two intensity 
curves corresponding to displacement in the field of view. The 
continuous line is the curve for the Fabry-Perot silvered plate, 
while the dotted curve is for plates of low reflecting power. 

Displacemcnr in Field of VieuJ 
Fig. 218 

Note in the one case the broad intervals of practically no 
intensity and contrast with the other in which the intensity falls 
off slowly, leaving comparatively small dark intervals. 

Lines due to a second ray of slightly different wave length 
from that for which the second curve is drawn would produce 
crests over the dark intervals and leave the field nearly uniformly 
bright, and it would not be possible to distinguish the separate 

The determination of the Ratio, — , for an Electron by means of the 


Zeeman Effect 

When a monochromatic source of light is placed in a magnetic 
field, and rays are received in a direction parallel to the direction 
of the field, it is found that the normal frequency is changed, 
and two or more lines appear symmetrically displaced from the 
usual position in each direction. 

The light in each case is circularly polarized ; the component 
of higher frequency is polarized in the opposite direction of 
rotation to that for the lower. 

Reference for the theory of this phenomenon should be made 
to " The Electron Theory of Matter" (O. W. Richardson). 

If H denotes the intensity of the magnetic field, and its direction 

is parallel to that of the light, the increase and decrease in 

1 eH. 
frequency of the two components is of magnitude — • — > 

and H is measured in ' gausses,' the name given to the electro- 
magnetic unit of magnetic field. 
In the experiment we measure the difference in wave length 


of the two components, i.e. we measure <$x, corresponding to a 
change of frequency, 

*» = — --H. 


Now &v = -g <5x, (numerically), 

since c = v\ 

1 X 2 e 

so that we measure <5x = — — H . — c m 

tt e c 5\ , ' 

Hence: -.^.g.- (l2) 

is measured in electromagnetic units. 

An electromagnet with adjustable pole pieces and taking a 
current of three amperes gives good results with an ordinary 
vacuum tube. 

A hole is drilled in the pole pieces so that the source may 
be observed along the direction of the magnetic field. 

The field is measured by means of a fluxmeter, by which the 
total flux through an exploring coil placed between the pole 
pieces is measured. This will be read directly on the fluxmeter. 
Suppose it is B, Maxwell's. If A is the effective area of the 
coil and H the field, on placing the coil into the position, where 

H is required, the flux is HA, so that H = -^, and H is in gausses 

(see experiment on Grassot Fluxmeter, p. 482.) 

We have now to determine <5X by means of the Lummer plate. 
First align the apparatus and drilled hole as nearly as possible. 
Place a sodium flame on the far side of the hole, and obtain the 
shorter sodium lines as sharply as possible with the drum set at 

Place a helium tube or other convenient source in position 
between the pole pieces, and obtain the yellow line. The yellow 
line .was used in an experiment and satisfactory photographs 
were obtained, but it would be preferable to use the blue line 
so that the exposure is not long, and ordinary photographic 
plates (special rapid) may be used. We shall, in what follows, 
consider the yellow line to have been employed, and the wave 
length is then 5875-6 A. 

Adjustments will be carried out as described above, and the 
best possible definition and illumination obtained. 

On applying the magnetic field it will be observed that the 
lines broaden, and in some cases actual separation will occur, 
as in fig. 219. 


The state of polarization may be tested by intercepting the 
light just before it enters the object glass of the telescope by a 
quarter- wave plate. 

The plate must, of course, be suitable for the particular wave- 
length concerned — the usual plate met with in laboratories will 
be suitable to examine the yellow light. This part of the experi- 
ment is introduced so that the student may take the opportunity 
of verifying the character of the vibrations in the lines. 

Fig. 219 

The quarter-wave plate reduces the circular vibrations to two 
linear vibrations at right angles to one another, and on examining 
these with a Nicol it will be found that when one line is cut 
out the other is present in the field. 

After carefully adjusting the apparatus so that the effect is 
seen by eye, attach the camera and photograph the spectrum 
without the magnetic field. By means of the rack and pinion 
raise the plate, put on the field and take a second photograph 
below the first. 

The plates when developed will give lines as in fig. 219. 

Separation of the lines is seen in the higher orders, e.g. at A 
and B. Let a denote the line when no field is applied to the 
source, and A the corresponding line with the field applied 
Measure the mean displacement of the components of A by 
means of a microscope, and measure the separation between the 
two successive orders a and b. 

Denote these distances by I and L respectively. 

We proceed to work out a formula, showing how <5x may be 
. / 
derived from the ratio j- 

By the equation for the Lummer-Gehrcke plate : 

2,\sd cos r = n\, (13) 

we see that in proceeding to a neighbouring order (n -f- 1), the 
angles of emergence and refraction are (*' -j- 51) and (r + <5R), 


where: 2d\i cos (y -f <5R) = {n + 1) x (14) 

From (13) and (14) we have by subtraction : 

— 2d tan r cos * <5I = x •••(15) 

In this step we use the relation : sin (*' + 61) = y. sin (r + <5R), 
and consequently cos idl = y. cos y<5R (16) 

There is no variation in n since the wave length remains 
constant ; we pass merely to a new angle of emergence. 

<5I represents the angle between the rays which produce two 
consecutive lines as a and b on the photographic plate (fig. 219). 

Now consider the difference in direction, 5i, for two lines in 
order, n, of wavelengths, x and x + <5X. , 

It is an angle, 8i, which corresponds to the displacement 
between the components of A. 

Since the plate is fixed in the sjfectrometer and the angles are 
small : 

8i I . . 

sT=L <M 

Referring once more to equation (13) we have in the order, n, 
a wave length, X, corresponding to a direction measured by *, 
and a wave length, x + 8\, corresponding to * + Si. 

By differentiating (13) we find : 

— 2d[i sin rdr = w<5x — 2d cos r -=- <5x (18) 

and from the equation, 

sin * = [t sin r, 

. . . du. 
fji cos rdr = cos 1 St — sm r j~ <5X (19) 

In this case it is necessary to take account of the variation 
of (a since the wave length changes. 

By eliminating 8r from (18) and (19) it is found that : 

— 2d tan r cos i 8i = (n 4- J <5x ; (20) 

\ cosy d\J v ' 

and on substituting for n from (13) : 

— 2d tan r cos i 8i — 2 ( - cos r — ^ \ 6\ (21) 

\X cosy d\J v ' 

Thus from (15), (17), and (21) : 

I 6i 2d/y. I d\x \ . , . 

=-= 7r= — ( —cos r • — - ) <5x (22) 

L 81 x\x cosy d\J K ' 

The value of y is the critical value for quartz, and it remains 
to determine j- for the particular wave length used in the 


experiment. The table below showing values of n corresponding 
to different values of x should be used to draw a graph from which 

~ may be obtained by measuring the slope at any point where 

it is required. 
From the equations (12) and (22) we find : 

2.TZC I 

W *L 








and the units are electromagnetic 

If possible take the ratio-y- for more than one order — several 
orders will usually show sufficient separation for this purpose. 





















1 -5410 





Measurement of Wave Length by Diffraction at a Straight Edge 

The theory of this experiment will be found in Schuster's 
" Theory of Optics " in Chapter V. 

In the first few sections of the chapter it is shown that on 
observing a straight edge illuminated by monochromatic light 
from a narrow slit a series of alternate bright and dark bands 
will be seen. 

If S denotes the slit (fig. 220), EF the straight edge fixed 
parallel to it, and PQ a plane perpendicular to SE, we 
shall have the series of bands along PQ. Let Q be the nth 
bright band from P, where P is the point of intersection of SE 
and the plane. 


Denote PQ by x n , then it can be shown that : 


Xn=s ^PH4n~i)(p-hq) 

In this formula p is the distance, EP, and q is equal to SE, 
the distance between source and slit. 

F c 


Fig. 220 

The bands are not regularly spaced as are interference fringes. 

To carry out the measurement set up in one of the stands of 
the optical bench a straight edge parallel to the slit which is 
illuminated by sodium light. First adjust the parallelism by eye, 
and finally make a slight rotation by the screw of the holder until 
the bands are most distinct ; then make the slit as narrow as is 
compatible with sufficient illumination. Observe the fringes 
with the travelling microscope, and measure the distance between 
the first or second bright band and one of the most distant 
that can be seen plainly. 

If this is the nth and the first observed is the mth we have : 

x n -x m = ^(£+g) |v^=l - V^t^i] . . . .(25) 

The student should take the opportunity of observing the 
diffraction fringes arising from the light passing a narrow wire, 
needle point and narrow slit 

In the case of the narrow wire diffraction bands unequally 
spaced will be seen outside the geometrical shadow. Within it 
a series of equally spaced bands will be observed. These may 
be described as interference fringes due to the two parts of the 
wave, one on* either side of the wire. The effects of these are 
equal to those of the two half period zones which lie at the 
edges of the wire so that they act like two sources at a small 
distance apart. 


Determination of the Wave Length of Light by means of a Plane 
Diffraction Grating 

A diffraction grating is made by ruling a large number of 
equidistant parallel straight lines on glass. The lines are ruled 
by a diamond point moved by an automatic dividing engine 
containing a very fine micrometer screw which moves sideways 
between each stroke. A photographic replica of a plate made 
in this way is often used in its place. 

The number of gratings to the inch is marked on the glass. 

In handling the grating do not touch the faces of the glass, 
hold it between thumb and finger by the edges. 

Adjust the collimator and telescope for parallel rays in the 
usual way and observe the direct image of the slit noting how 
it lies in the field of view. 

Set up the grating with its face normal to one side, EF, of the 
triangle formed by the levelling screws (cf. fig. 161). Throw an 
image of the slit into the telescope by reflection from one face 
of the grating and adjust the screws to bring it into the same 
part of the field as that occupied by the direct image. This 
makes the faces vertical. 

In order to set the grating at right angles to the rays adjust 
the collimator and telescope at right angles to each other, and 
turn the table until the slit is reflected on the cross-wires of the 
telescope. Then turn the table a further 45 . 

It now remains to tilt the grating so that the lines are parallel 
to the slit. 

View the first diffracted image, making the slit as narrow as 
is convenient, and adjust the screw D until the best image is 
obtained. The lines and slit are then parallel. 

Find the diffracted images on both sides of the line of direct 
vision. It will be easy to observe two orders, and if a bright 
light is used and an image of the brightest part of the flame 
thrown on the slit by a short focus lens the third may be seen also. 

If (a + b) is the width of the grating element the formula is : 

(a + b) sin = wX, 

for normal incidence, being the angle of diffraction, X the 
wave length of the incident light, and n the order of the spectrum. 

Obtain by taking half the angular distance between the 
corresponding images on each side. 

(a + b) is deduced from the number of rulings on the grating. 

Use any source of monochromatic light or light giving well- 
marked lines as that from a discharge tube containing hydrogen 
which gives three well-marked lines, red, green, and violet, known 
as C, F and Hy. 


The above formula is obtained by considering rays in pairs 
passing through adjacent clear spaces of the grating. 

For example, consider the two spaces AB and CD. 

We may consider the rays passing through them in pairs, 
taking together those rays which are symmetrically situated, as 
for example, QLQ 1 and TMT 1 . 

These reach the grating in the same phase, and the line ABCDE 
represents a section in the plane of the figure of the wave front 
incident on the grating. 

Fig. 221 

Consider a wave front, WW, at a later interval, and suppose 
that it makes an angle, 0, with the plane of the grating. 

Draw the rays AP 1 , BR 1 , etc., perpendicular to this wave front 
and making the angle, 0, with the normal, AO, to the grating. 
These rays are all received by a telescope and united in the 
focal plane. Their paths from the grating to the instrument 
differ, so that they reach the focal plane with different phases. 
This phase difference is due to the difference of path traversed 
after leaving the grating. 

Take the case of the two rays, AP 1 and CS 1 . Their path 
difference is AN, where N is the foot of the perpendicular from 
C on AP 1 . Any other pair of rays, e.g. LQ 1 and MT 1 , situated 
symmetrically in AB and CD have the same path difference. 
This holds for all pairs of rays symmetrically situated. Thus 
the phase difference on arrival in the apparatus that receives 
them is the same for each pair. 

We shall thus have reinforcement if AN is equal to a whole 
number of wave lengths. 

Now AN = AC sin 0. 


If a is the width of a space and b that of a line, 

AC = a + b, 
(a + b) is called a grating element. Thus a bright line occurs 
in a direction 0, provided that 

(a + b) sin = n\. 

The Plane Reflection Grating 

This reflection grating consists of a plain polished sheet of 
metal across which parallel lines are drawn very closely together 
as in the transmission grating. 

It is necessary to mount the polished face vertically and this 
is done as in the previous case. 

The light does not fall from the collimator normally on the 
grating, but at a measured angle, *. 

In order to measure i, the grating is first set with the polished 
face normal to the rays from the collimator which, together with 
the telescope, has been adjusted for parallel rays. 

The table carrying the grating is turned through a definite 
angle, say io°, and is then fixed. 

Suppose rays from any direction make with the normal to 
the grating and are received in the telescope. 

Consider any three rays, MA, NB, and OC (fig. 222), incident 
on the grating element ABC. AB is polished while BC is the 
position of the line where the polish is scratched and where the 
incident rays are absorbed. 

A B C D 

Fig. 222 
The path difference between the extreme rays MAQ and OCS is 
AF—CE, where the dotted lines, AE and CF, denote incident 
and reflected wave fronts respectively. If AB = a, BC = b, this 
difference is equal to 

(a + b) (sin 0~sin *). 
It is possible to divide up the bundle of rays falling on AB 
and CD into pairs, one from each bundle, having this same 


difference of path. In order to do this it is only necessary to 
choose rays occupying the same relative positions in the two 

If this path difference is a whole numbsr of wave lengths, 
and the emergent parallel rays AQ, BR, CS, etc., are brought to 
a focus by a telescope, we shall get a bright image due to reinforce- 
ment of the rays. We thus have : 

(a + b) (sin 0*~sin i) = n\, 
where n may have the values 1, 2, 3, etc. The corresponding 
values of 0, say, B lt 2 , 3 , etc., give the directions in which 
the different orders of diffracted images are seen. 

If the observation is made from a direction on the same side 
of the normal as i, the formula is : 

(a + b) (sin i -f sin 0) = nx. 

Obtain first the reading of the scale marking the position of 
the telescope when it is directed towards the slit with no grating 
intervening. As above, set the normal to the grating at some 
definite angle, *, to the incident rays. Move the telescope round 
to receive light in a direction corresponding to = i on the 
opposite side of the normal to *. This corresponds to order zero 
— it is the position where the ordinary reflected beam is received. 
Now move round to the next position where a distinct image 
can be found. This will correspond to n = 1. Carry out this 
process as long as it is possible to observe images at all. The 
higher orders get fainter but resolve the lines more than the 

The number of lines to the inch is given for the particular 
grating, so that it is possible to deduce (a + 6), which must be 
expressed in centimetres. From the values of *, and n, it is 
then possible to evaluate X. It should be possible to obtain 
separation of the sodium lines in the second order, and the 
wave length for each constituent should be calculated. 

Use three or four different values of i and compare the value 
of x obtained, finally take the mean of the series. 

Resolving Power of a Telescope 


Let a parallel beam of monochromatic light fall on a slit, AB, 
and let us examine the intensity of the light along a direction 
CE, inclined at an angle, 0, to the normal to the slit. Let the 
light in this direction be brought to a focu$» at E. From A draw 
ADF perpendicular to the beam so that z. BAF is 0. The rays 
issuing from the slit on arrival at E will have different phases 
on account of the varying length of their paths from AB to E. 
We may measure these differences by such lengths as PM, varying 
in amount from zero at A to a maximum of BF for the extreme 
ray from B. 


Divide the slit, AB, into a very large number of equal parts, 
such as PQ. If we make each of these small enough, all the rays 
in the element of which PQ is a section, may be supposed to be 
in the same phase on arrival at E. Suppose that the rays from 
each small element differ in phase by an amount, 3, from the rays 
of the adjacent element. Each element will be supposed to 
contribute an amplitude, A, to the whole beam so that if there 
were n slits, and all had the same phase on reaching E, the 
amplitude would be«A. 

Let the disturbance at A be denoted by ' A sin pt. 

This is an expression which represents the displacement in 
any simple harmonic vibration, and rays of light afford an example 

Fig. 223 

of this type of vibration. The rays from the next element, 
differing in phase by 6 must be represented by A sin (pt + <5), 
from the next by A sin (pt + 28), and so on for all the elements 
of number n. 

Thus the total effect at E is obtained by adding up these 
separate elements. 

Let T measure this effect so that : 

T = A {sinpt + sin (pt + d) + ... + sin (pt + n -1 <5)}. 

The maximum phase difference is measured by 2n -— radians. 


Let <p denote this phase difference ; it is also measured by n — 1 8. 
The above series for T can be expressed by the more convenient 
formula : 

T = 

A sin (pt + - — — 3 J sin - <5 

. 6 


as is shown in textbooks on Trigonometry. 
When 6 is very small, sin \d = £<5 ; 

n A sm — 

= «Asta l£sin( ^ + W) 

in which |<5 is neglected because of its smallness. 

When 9 is zero there is no difference of phase, and the direction 
CE is normal to AB. In that case the amplitude at E is nk, 
and we denote this by B. Thus the amplitude in any direction, 

EF, is , *? , the factor sin {pt + \<p) denoting the oscillating 

character of the disturbance. 
The intensity is proportional to the square of the amplitude, 

or the intensity at E is measured by : 4_ — ' sm t L. 

rrw - • 1- sin 2 £? . 

This is a maximum when — ^~ is a maximum. 

9 2 
If this function be examined for its maxima and minima by 
means of the differential coefficient, it will be found that the 

maxima occur at points where the values of - are : 

o, i-437r, 2-4675, 3-4775, etc., 
and the minima at 

rc> 27r, 375, 47c, etc. 
At the latter values the intensity is zero. 
The graph drawn in fig. 224 shows the relation between intensity 
and values of \<p. 

When the light is examined at H the intensity corresponds to 
the ordinate at I, and on moving round we come to a place, E, 
say,where there is darkness corresponding to the zero ordinate at 75. 
In this case : 

1 1 BF 

i.e. BF^= X. 

Thus a minimum occurs in the direction in which BF == X. 
Now the angular distance between CE and CH is the same 

as angle BAF, and since this is small it is measured by 

or x 

(breadth of incident beam). 



^ M 


Fig. 224 

If another beam were incident on AB in the direction along 
CE, there would be a maximum for this beam at E, and it would 
overlap the minimum of the former. 

The intensity curves might then be represented on the same 
diagram as in fig. 225. The curve on the left represents the 
intensity curve for the second direction while the upper dotted 
curve shows the two compounded. 

Fig. 225 

The resultant has two pronounced maxima with an appreciable 
dip between. It will thus be possible by the aid of this falling 
off in intensity between two bright regions to distinguish the 
two beams or they will be 'resolved.' Moreover, the angle 
between the two directions is ECH or X ~ (width of beam). This 
is taken as a limiting case, and the resolving power is measured 
by this ratio. If the angle between the incident beams is less 
than this the two maxima cease to be distinguishable as two 
and blend into one. 

In order to verify this theory a telescope is fitted with an 
adjustable slit which is placed as close as possible to the object 


glass. The width of the slit is carefully measured by means of 
a micrometer microscope. 

A suitable object for this experiment may be made by coating 
a sheet of plane glass with tin-foil and cutting two fine parallel 
lines in the foil with a razor blade at a distance of two or three 
millimetres apart. When these are illuminated with a sodium 
flame they provide two bright slit-sources. This object should 
be placed at different distances from the object glass of the 
telescope, the aperture of which may be varied by means of an 
adjustable slit placed immediately in front. 

A certain minimum width of this slit will be found for which 
the two lines appear as separate lines. This width varies with 
the distance from the object glass to the two slits, and for smaller 
widths the lines appear as one. 

A table is made of the minimum widths of the slit and the 
corresponding distances. 

The angle subtended by the two fine lines at the object glass 

is measured by== > where d is the width of the slit and D the distance. 

The theory above described shows that the value of this angle 

is-> where x is the wave length of the light, and a is the width of 

the aperture. 

The object of the experiment is to compare the theoretical 

and practical resolving powers, the former determined by - and 

the latter by ~* 

An examination of the practical resolving powers measured in 
the Wheatstone Laboratory of King's College, London, during 
the past session has shown that the values obtained were about 
20 per cent, greater than the theoretical value. 

Repeat the experiment with two point sources and circular 
apertures of different diameters. 

In this case the field of view contains a bright small central 
circle with concentric alternate bright and dark rings. If two 
sources close together are such that the bright centre due to 
one falls on the first dark ring of the other, the two sources just 
cease to be distinguishable as separate. 

Sir G. Airy has shown that if d is the diameter of the aperture, 
the angle subtended when separation ceases, 

e = i-22-j. 

The point sources should be two fine holes in a sheet of tin. 

They should be illuminated by monochromatic light and placed 

at such a distance from the object glass that they just cease 

to be seen as two holes. 



Place the telescope at different distances from the holes and 
focus it on them. Adjust the aperture diameter until the holes 
just cease to be distinguishable as two separate sources of light. 
Measure the distance between the centres of the holes by means 
of a micrometer microscope, and the distance between the plane 
of the holes and the aperture by a metre rule. From these 
measurements deduce 6 and compare it with the theoretical value, 

x • v 

i'22-j, in each case. 

Polarization by Reflection. Verification of Brewster's Law 

When light is reflected from surfaces the reflected beam is 
partially polarized, that is to say, that the transverse vibrations 
constituting the light have, on the whole, a greater component 
along a particular direction than in any other. 

Ordinary light is supposed to consist of a transverse vibration 
which changes its direction in space, though of course always 
in the wave front, so rapidly that on the average in any appreci- 
able interval of time the component in one direction is the same 
as that in any other. 

The reflected light has lost this property and is polarized so 
that it has a greater component in one direction. This direction 
is normal to the plane of incidence. 

The transmitted light has a greater component in the plane 
of incidence. 




Fig. 226 

In the diagram (fig. 226), a ray, SA, is represented as being 
reflected at a glass sheet, MM, so that AB is polarized. In 
order to test the polarization the ray is received in an analyser, 
which in some instruments consists of a Nicol prism. This- is 
a prism of Iceland spar which is cut into two along a diagonal 
plane and cemented together with Canada balsam. Iceland 
spar has the property of dividing a ray of light into two rays 
refracted in different directions and one polarized perpendicularly 
to the other. The layer of Canada balsam serves to reflect one 


of these rays to the side of the prism where it is absorbed by. 
the blackened walls of a case, and the other ray is transmitted. 

In this way the emergent beam is made to consist of light 
vibrations all in one direction. Incident light with its vibrations 
in this direction passes through the Nicol, while if the vibrations 
are perpendicular to this direction the light is unable to get 
through. Light with vibrations in any intermediate direction 
has only the components parallel to the direction of trans- 
mission passed on. 

The vibrations transmitted are parallel to the shorter diagonal 
of the end of the prism. 

On examining AB with this prism it will be found that as 
the prism is rotated there is a change of intensity in the trans- 
mitted light. This means that AB consists of vibrations with 
the components in one direction greater than in another or it is 
partially polarized. 

On altering the inclination of MM the alternations can be 
varied in extent, and in one position the change between 
brightness and darkness is a maximum. The angle of incidence 
when this occurs should be noted by the help of the scale of 
angles attached to MM. Theoretically the light should be 
completely polarized in one position, so that for one particular 


Fig. 227 

setting of the Nicol darkness should be complete. In practice 
it will be found that the light is not quite all cut out, though 
with care a position will be found when, this is very nearly true. 
The apparatus is represented diagrammatically in fig. 227. 
Brewster's law is that for complete polarization : 

tan t = {*, 


where * is the angle of incidence and ^ is the refractive index 
of the reflecting material. 

This law should be verified. It will be found best to allow 
light from a window on the opposite side of the laboratory to 
fall on MM, and to adjust for the maximum effect, and afterwards 
to set up a sodium flame as at S, and make the exact adjustment. 

In another form of apparatus the Nicol is replaced by a second 
glass plate which can be rotated about the direction AB as well 
as inclined at different angles to the horizontal. 

When the mirrors are parallel the polarized light in AB is 
readily reflected by the second mirror, while on rotating it through 
a right angle from this position about AB this polarized light 
will not be reflected. 

We shall thus obtain alternations in intensity on rotating the 
mirror about the vertical axis, and there will be a maximum 
effect for a particular inclination of MM, in which case tan i. =\j. 

Rotation of the Plane of Polarization. Laurent's Saccharimeter 

The essential parts of the saccharimeter are two Nicol prisms, 
N x and N 2 , illustrated in fig. 228, one of which serves to polarize 
a beam of light passing through it while the other analyses the 
transmitted beam and detects its plane of polarization. These 
Nicols are spoken of, respectively, as the polarizer and analyser. 


» t^y o 

L N, H 

Fig. 228 


N 2 

When N x has reduced the light vibrations to a particular 
direction, viz. parallel to the short diagonal at the end of the 
prism, all the light transmitted by N 4 can pass through N 2 if 
N 2 is oriented exactly in the same way as Nj, i.e. if its shorter 
diagonal lies parallel to that of N lf and its length lies parallel 
to that of N ; . We are here neglecting the diminution of intensity 
due to absorption, which always goes on, since actual bodies are 
not perfectly transparent. We mean that no light is cut out 
in this case on account of polarizing effects of N 2 . 

In this position the Nicols are said to be parallel. If, however, 
N t is turned from this position through a right angle no light 
from N! can get through N 2 , since N 2 is now so oriented that 
the light vibrations falling on it are in a direction perpendicular 
to its short diagonal, and such vibrations are not transmitted. 

In this position the Nicols are said to be crossed. 

Certain substances like quartz, and solutions like that of sugar, 


possess tiie property of rotating the plane of vibration of light 
as it passes through them, so that if N x and N 2 are crossed when 
the active substance is not placed between them (in which case 
no light will get through N 8 ),on inserting the active material, 
on account of the change in direction of the vibration, some 
light will pass through N 2 . 

It is found that a rotation of N 2 in one direction or the other 
will bring it into a position when the light is once more stopped. 
This shows that the light is still polarized, but its vibrations 
have changed direction in traversing the medium. We ought, 
therefore, to be able to measure the amount of this rotation by 
measuring the angle through which N 2 is turned ; but, unfor- 
tunately, N 2 can be turned through an appreciable angle when 
the light is cut out without any apparent return of the light. 
This lack of sensitivity is overcome in the saccharimeter by a 
special device. 

Just in front of the polarizing Nicol on the side towards the 
analyser is placed a semicircular sheet of quartz cut parallel 
to the optic axis. The complete circle is made up by a semi- 
circle of glass of such thickness that it absorbs the same amount 
of light as the quartz. The position of this circle is at H, and 
it covers the open end of N x completely. 

F A 

O ~ Q 

Fig. 229 

When the light falls on the quartz it is separated into two 
components, polarized normally to one another, which travel 
through the quartz with different velocities. Let one component 
be represented by OI and the other by OQ just as the light reaches 
the quartz plate, these are the components of a vibration along 
OR ; It is here supposed that the ether particle is at O but 
just moving in the direction OR, so that its component directions 
are OQ and OI. As the disturbance passes through the plate 


there will be a gradual change of phase between these components 
on account of the differing velocities of transmission. After a 
time the disturbance will reach a point in the plate where one 
component displacement is along 01 while the other component 
is along OQ 1 . These combine to give a displacement OR 1 . 

The quartz plate is cut so that, as the disturbance just leaves 
the plate on the other side, this difference of phase exists between 
the components. The difference is one-half of a period, and the 
plate is called a half-wave plate or half shade. 

Of course the light which traverses the glass side proceeds 
undisturbed, and its oscillations are still along the direction, ED, 
shown in the upper part of the diagram, parallel to OR. 

If N 2 receives this light when its short diagonal is at right 
angles to OR 1 this component is not transmitted, and no light 
from the quartz side gets into the eye, while part of the com- 
ponent OR gets through and the glass side appears illuminated. 
Generally, light passes in from both sides of the plate, but both 
sides are not equally illuminated. When both sides present the 
same illumination the principal plane of the Nicol is either along 
AB or normal to it, for it is clear that in either of these positions 
the components transmitted are the same for both sides. 

It happens that the eye can readily detect a change from 
the equality of illumination in both halves of the field, parti- 
cularly if both halves are equally dark, i.e. when the Nicol is so 
placed that the smaller components are transmitted. 

If the Nicol is set for equal iUumination on both sides, and 
an active substance is interposed, it will be necessary to rotate 
the Nicol to find once more the position of equal intensities. 
The amount of rotation measures the angle of rotation of the 
plane of polarization. 


Fig. 230 

Another common method of bringing about an increase in 
sensitivity is to use the biquartz. This consists of two semi- 
circular discs of quartz fitted together to form a complete circle. 
One of these rotates the plane of polarization of the incident light 
in a clockwise, and the other in a counter-clockwise direction. 
The amount of rotation per unit thickness varies with the colour 


of the light. For a particular thickness the rays from a sodium 
flame will be turned in opposite directions through a right angle, 
so that if the short diagonal of the Nicol lies parallel to this 
direction the yellow rays get through, and if the diagonal is 
perpendicular to this direction these rays are cut out. 

When white light is used it is robbed of the yellow constituent 
when the Nicol lies in this latter position, and the colour observed 
is greyish and is called the tint of passage. 

It is easy to detect a slight change from this uniform colour, 
for an appreciable change takes place to a partly blue and partly 
red field, one colour belonging to each side. 

Let LMNS denote one end of the analysing Nicol, and let UV 
denote the direction of vibration of the light. 

This light will be cut out if MS— the shorter diagonal— lies 
normally to UV. A small rotation of MS counter-clockwise will 
bring it into a position to cut out the light and so will a larger 
rotation in the opposite direction, the sum of these rotations 
being 180 . 

It is thus not easy to decide which way the plane has been 
turned. But if two lengths of the rotating substance be used, 
one slightly longer than the other, the rotation for the longer 
must be greater than for the shorter. 

The direction of rotation of MS which shows a larger angle 
in the case of the longer is the direction in which the rotation 
has taken place. 

Tubes of glass with carefully worked end-pieces are used to 
carry the solution to be examined. The ends are held in position 
by metal caps screwed against them. It is necessary to have 
rubber washers between the glass ends and the tube to avoid 
strain when screwing up ; for a strained end will produce rotation. 

Find the amount of rotation for a solution of sugar in water 
and deduce its specific rotation. This quantity is denned to be 
the amount of rotation produced by one decimetre of solution 
divided by the weight of dissolved substance In unit volume. 

Let w grammes be dissolved in ioo ex. and suppose a length, 
/ cms., produces rotation, 0. The specific rotation is : 

6 w e 

iot -. = iooo T' 

I ioo lie 

Repeat for various strengths of solution, and for different 
lengths of tubes. 

When several tubes are obtainable it is interesting to observe 
the effect of causing transmission through different lengths of 
a solution of a particular strength. 

By this means it may be verified that the amount of rotation 


is proportional to the distance traversed by the light in the 

Another instructive experiment is to make solutions of different 
known concentrations, which may be measured by the number 
of grammes of substance dissolved in ioo c. c. of solvent, and to 
measure the amount of rotation in traversing a particular distance 
through the solution. A curve showing the relation between 
the rotation and concentration should be plotted. 

The Lippich Polarizing System 

In modern polarimeters the half shade is replaced by a more 
convenient method of dividing the field. In the most recent 
polarimeter the field is divided into three parts (fig. 231), the 
two outer similarly illuminated for all positions of the analysing 
Nicol and the central portion which may be differently illumin- 
ated from the neighbouring regions and which has to be matched 
with the outer parts of the field. 

Fig. 231 

l n l 



1 j 

Fig. 232 

The mode of producing the divided field is illustrated in 
fig. 232. N is the polarizing Nicol and LL are two small Nicols 
fixed in position and mounted in a brass cylinder in front of N. 
The directions of vibration of the light emerging from the 
cylinder and falling on the optically active substance are repre- 
sented by the arrows, I, n, I. The central portion passes through 
the Nicol, N, only, while the outer parts pass through N and L. 

When the analyser, M, is turned so that it transmits vibrations 
along a direction bisecting an angle between n and I the whole 
field is uniformly illuminated. For all other positions of M the 
field is not uniform. 

Thus, to measure the amount of rotation of any substance 
placed at A, the analyser is first put into the position correspond- 
ing to uniformity of field. The substance to be examined is 
then put into position and M again rotated until the field is 
once more uniform. The angle of rotation measures the rotation 
due to the active substance. 


The optical system by which the field is examined is not shown 
in the diagram. It lies to the right of M and is focussed on the 
plane through the right-hand ends of the Nicols, LL. This 
system has the advantage that it is suitable for the examination 
of all wave lengths, whereas the half shade has to be constructed 
for one wave length only. Fig. 231 shows at OCO how the field 
is divided into three parts. 

The Half Shadow Angle 

The various devices employed to enable accurate observations 
to be made in polarimetry, which have been described, produce 
two beams of polarized light with vibrations in directions inclined 
to one another. 

In the Lippich system we have denoted the two directions 
by I and n. 

In fig. 233 these directions are denoted by B/ and Bw respec- 
tively, and the angle between them is 20. This angle is called 
the ' half shadow angle,' and the magnitude of this angle has 
an important bearing on the question of sensitivity. 

Fig. 233 

Suppose that BC bisects the half shadow angle, and that DBE 
denotes the direction of vibration of the light which traverses 
the analyser. 

When this lies at right angles to BC the intensities of the two 
beams are equal as seen through the analyser, for the components 
of the displacements transmitted, viz. BF and BG, are equal, 
and the intensities are in the ratio : 

BF 2 : BG 2 . 
Suppose the analyser is turned through an angle, a. 
Then the transmitted components are : 

BZ cos ZBE 1 and Bw cos nED 1 , 
or BZ sin (0 — a) and Bn sin (0 + a). 

Thus the ratio of the intensities is : 

sin 2 (0 — a) : sin 2 (0 + a). 

In photometric work it is assumed that the eye can detect a 
difference of intensity of one per cent. 
Thus, if we regard as a given angle, we may say that the 


change in setting of the analyser of the amount a will just be 
detected when 

sin 2 (0 — a) = «99 sin 2 (0 + a). 

From this equation it follows that when a is a small angle, 
so that we may write : 

sin a = a, and cos a = I, 
a = -0025 tan 0. 

If a has a small value the apparatus is sensitive, and it would 
appear that the sensitivity is improved by making as small 
as possible. 

But as gets small difficulties arise on account of the fact that 
the light is never plane polarized, it is always elliptically polarized 
in practice. 

may not be indefinitely diminished. 

Polarimeters are usually fitted with a small movable arm 
projecting from the tube which carries the polarizer. This arm 
carries an index mark which moves over a scale. By means 
of it the polarizing Nicol can be rotated so that the half shadow 
angle can be adjusted within limits. A modern form of the 
apparatus is illustrated in fig. 234. 

The sensitivity is thus to some extent under the control of 
the observer, who will discover as he becomes familiar with his 
instrument the best adjustment for sensitivity which suits him. 

The reader may be referred for a more detailed and complete 
account of all these questions to the article on " Polarimetry " 
in the " Dictionary of Applied Physics." 

Soleil's Compensator 

Sometimes the saccharimeter is fitted with a piece of apparatus 
consisting of two quartz wedges. This is known as Soleil's 
Compensator. Fig. 235 illustrates the apparatus. The wedges 
are ABC and DEF, and these are mounted in metal holders 
which can be moved by means of a rack and pinion, so that the 
wedges are translated in either direction parallel to AC or DE. 

Thus if a ray of light is passed perpendicularly to AC it is 
possible to place varying thicknesses of quartz in its path. The 
quartz wedges are cut so that the optic axis lies perpendicular 
to AC and DE, and polarized light passing through them suffers 
rotation of its plane of polarization. 

The amount of rotation can be varied by moving the wedges 
by means of the rack and pinion. 

The quartz wedge is placed just in front of the analyser. 

An index mark moves along a scale as the wedges are displaced, 
so that a record can be made corresponding to each thickness 
of quartz interposed. 

i'aae 378 


The analyser is first rotated until with the quartz wedges 
occupying a convenient zero position the field is equally dark 
on both sides, supposing that the polarimeter is fitted with a half 
shade. The quartz wedges are then moved a small amount by 
the rack and pinion, and the amount of rotation of N 2 necessary 
to restore the uniformly dark field of view is recorded. 

Fig. 235 

By making a number of observations a curve can be plotted 
which shows the rotation corresponding to the various dispositions 
of the wedges. 

When light has undergone a rotation before passing through 
the wedges, this rotation may be counteracted by interposing 
the correct thickness of quartz. 

Thus suppose the Nicols and quartz occupy the zero position 
described above, and that an optically active material is inter- 
posed, the rotation resulting may be counteracted by displacing 
the quartz wedges either so as to increase the thickness traversed 
or to diminish it. When the appearance in the analyser is the 
same as that of the zero position, we know that the wedges have 
caused a rotation equal in magnitude, but opposite in direction 
to that of the active substance. 

Thus by observing the record opposite the index mark we can 
deduce the amount of rotation due to the substance. 

In plotting the graph, rotations in one direction will lie on 
one side of the origin up to 180 °, while those in the other direction 
will lie on the other side up to 180 . 




The light emitted from a small source is absorbed very little by 
the air through which it passes, so that we may say that any 
surface, surrounding the source completely, will receive the same 
total amount of light. 

Let this total amount be denoted by M. 

= luii 

Fig. 236 

Imagine a cone with its apex at a small source of light and let 
its solid angle be w. All surfaces receive the same amount 
of light on the parts lying within this cone. 

If w is small, say 6w, the cone may be regarded as denning a 
particular direction and the intensity will be regarded as the same 
for all rays within this cone. If L denote the total amount of 
light emitted within this cone per second, we write : L = Kdw. 

This equation defines K, which is sometimes called the 
'candle power ' of the source. 

In general, K is dependent upon the direction, but when K is 
the same for all directions 

M = 4ttK, 

where M is the total amount of light emitted per second by the 

Let the small cone cut a surface in the element, <5S, and 
let the mean direction of this cone make an angle, 0, with the 
normal to &S. 


Then aS = 


r 2 8w 


, T K cos ^ 

and L = — -—- — <5S. 

r 2 

The intensity of illumination is defined to be the amount of 
light falling on unit area per second, 

i e I — — — ^ cos ® 

~ ss 7* 

The intensity, I, varies inversely as r* and directly as cos 0. 
When I is the same for two similar surfaces they appear to the 
eye to be equally bright, and on this principle the use of photo- 
meters depends. 

The efficiency of a source of light may be denned as the ratio 
of K to the material or energy consumed per second. 

The average candle power, in the case when K is not constant, 
divided by the amount consumed per unit time is called the mean 
spherical efficiency. This term is employed because the average 


value of K is denned as — , and this denotes the average amount 

of light falling on 1 sq. cm. of a unit sphere placed with its centre 
at the source. 
The efficiency of a candle is : 

K -^- weight of wax consumed per second, 
of a gas flame : 

K -r- cubic feet of gas consumed per minute, 
and of an electric lamp : 

K 4- watts supplied to it. 
The watts are measured by the product, volts x amperes, one 
watt denoting the rate of working when a current of one ampere 
falls through a potential difference of one volt. 

The unit in which K is measured is the candle power of a 
standard candle in a horizontal direction when the flame is 
50 mms. high. 

If another source is used and compared with the standard the 
two are adjusted to the same height and placed at such distances, 
r 2 , and r it from a conveniently placed screen that each makes it 
appear equally bright. In this case by what has been said above 
concerning intensity we have : 

t _ Kjcos T K 2 cos 
Ii - -JT- - I. - -jt— 


since the angles are the same, 
power its value is unity, and, 

If K, is the standard candle 

It is assumed that the student is familiar with the simple forms 
of photometer such as Rumford's and Bunsen's. It is difficult 
to make accurate comparisons of the illuminating powers of 
sources with these instruments. 

We shall be concerned chiefly with the more accurate types of 
instruments in this chapter. Even with these care and practice 
are required, but a skilled observer can obtain accurate results. 

The Efficiency of Sources of Light 

Take a gas-burner provided with an indicator registering the 
quantity of gas supplied, and adjust the flame until it stands at 
the same level as a candle flame and the grease spot of a Bunsen 

The candle may be taken as the standard with the value of K 
unity, and it must be shielded from draughts and must burn 

Determine K for the flame corresponding to different rates of 
supply of gas, by adjusting the distances between the sources, 
until the photometer screen appears alike on both sides. 

Plot a curve, showing the relation between the efficiency and 
the supply per hour. 

It will be found that the efficiency increases with the supply 
up to a maximum and then diminishes. 

It is most economical to adjust the supply to the value 
appropriate to the maximum. 


r v " 







Fig. 237 

We may similarly measure the efficiency of an electric lamp. 
Arrange the lamp, L, on the same level as the grease spot and 
candle flame, as before, and measure its candle power when the 
current is supplied at different voltages. 

Fig. 237 illustrates the arrangement of apparatus. 

V is a voltmeter joined to the terminals of L, and A measures 
the current in amperes. The resistance, R, is adjustable and is 



used to vary the current supplied to the lamp.. The power is 
obtained by connexion to the main through a plug. A con- 
venient voltage is 100 volts. 

Care must be taken not to short-circuit the mains. 

Plot a curve, showing the variation of candle power, with 
rate of supply of energy as measured in watts. 

The Flicker Photometer 

One form of this instrument is illustrated in figs. 238 and 239. 
The essential part consists of a white wheel, W, of which the edge 
is about 1 cm. wide, and is cut to the shape of a ridge running 
spirally. This is shown at RR, in fig. 239. The wheel is mounted 
in a box, black on the inside, provided with a rod, so that it may 
be supported in the carrier of an optical bench. 




Fig. 238 

The wheel is provided with a central axis and is rotated by a 
spring within the box, which is wound up by the key, K. 

The two sources to be compared are placed one on each side 
of the apparatus, as at S! and S 2 . These are carried in stands 
on the bench, so that the distances from the wheel are measured 
accurately and their heights are adjusted with the aid of two 
lenses fixed on the box. 

The small doors EF and GH, are closed, and the lenses used to 
throw an image of the sources on marked points, P 1 and P a , on 
the doors. When the images fall on these points the heights of 
the sources are correctly adjusted. 


The edge of the wheel is observed by means of an eyepiece, J, 
and usually unequal parts of the edge are seen on the two sides. 
When the wheel rotates the widths of these vary, and unless the 
illumination is equal on both sides a flickering effect is observed. 
If the illumination is the same on each side the ridge character 

S x and S 2 are moved to or from the apparatus until the flicker 
effect ceases, when the illuminations are equal, and if I x and I 2 
represent the illuminating powers of the sources, and r x and r 2> 
their distances from the wheel, we have : 


The distances should be measured from the middle of the wheel. 
It is not easy to decide when the "flicker ceases, and the degree of 
accuracy obtainable is not very high. 

Compare in this way a standard candle and a lamp. 

The Lummer-Brodhun Photometer 

This photometer is one of the most accurate. It is illustrated 
in fig. 240, and consists of a box, LMNO, containing the prisms, 
P lf P„A, and B, by means of which light is reflected and trans- 
mitted into the telescope, T, placed at 45 to the sides of the box, 
with its object glass in one corner. 

The two sources to be compared are at S x and S 2 , from which 
light falls on the slab, DD, which consists of magnesium car- 
bonate. From the diffuse sides of the slab rays are scattered and 
absorbed by the sides of the box, except those that cut the 
sides of V x and P 2 normally. These are reflected by the hypot- 
enuse faces into the two right-angled prisms, A and B. 

These prisms are the principal part of the apparatus. 

The hypotenuse of A is rounded off, except for a circular 
central portion which is placed in optical contact with the 



iarger face of B. The reason for this is, that rays of light may 
pass from A to B at this junction just as if the prisms formed one 
solid medium. Rays falling on other parts of the hypotenuse 
faces are totally reflected. 

In this way rays from P x pass on through B, forming the central 
bundle of rays in the beam emerging from the right and entering 
T. The rays outside this circle are reflected and are absorbed by 


Fig. 240 

the sides of the box. In the same way, rays from P 2 are 
reflected outside the circle, while those falling on the circle are 
transmitted. Thus the field of view of the telescope is illuminated 
by a central circle of rays, originating at S 2 , while the outer rays 
come from S x . Generally these two parts will be of different 
brightness, and on moving S x or S 2 , the two parts may be made 
equally bright. The eye can judge this easily and readily 
appreciate a slight deviation from equality. It is on this fact 
that the sensitiveness depends. 

From DD both sets of rays follow similar paths and light is 
absorbed equally. When the field of T is uniform, the slab is 


illuminated equally on both sides. If Lj and L 2 denote the 
illuminating powers respectively, and d x and d 2 denote the 
respective distances from DD, we have : 

L x L a 

df d 


Thus any two sources may be compared with accuracy. 

Compare two sources as in the last experiment. 

In order to eliminate errors arising from inequality of the 
reflecting power of the two surfaces of the slab and a possible 
difference in optical paths, DP X A and DP 2 B, it is usual to mount 
the box, LNOM, on a central horizontal axis lying perpendicu- 
larly to SiSa. Readings are first taken with the telescope as 
in the figure, and then with the box turned through two right 
angles so that the telescope lies on the left. 

The Nutting Photometer and its use for the Determination of the 
Absorption of a Solution 

This form of photometer is the most accurate instrument of 
its kind. It is based upon principles described by P. G. Nutting, 
but has been modified by Messrs. Hilger and Co., whose apparatus 
is here described. 

It is used in combination with the Constant Deviation 
Spectrometer (pp. 341 to 345), and the apparatus is illustrated 
in fig. 241, in the position ready for use. 

The essential features of the photometer are illustrated ki 
fig. 242. 

The box, A, is of aluminium, blackened inside, provided with 
two small windows, Q and Q 1 , by means of which light is admitted 
along the two directions, QR and Q X S. 

Light from a suitable source is deviated by two prisms, P and 
P 1 , carried in the plate, C. On the plate a number is inscribed, 
indicating the distance that the source must be placed from it, 
in order that the necessary deviation may be produced. This 
distance is usually 19 cms. 

Behind the window, Q 1 , lies a Nicol prism, Nj, which polarizes 
the light entering at Q 1 ; but behind Q in the form of apparatus 
shown no Nicol prism is placed. 

In another form a Nicol lies behind Q also, and its purpose is 
to counteract as much as possible the elliptic polarization that 
occurs on reflection at the surfaces marked R and S. It is 
found that this is reduced to a minimum by a particular orienta- 
tion of the Nicol. 

The prism, RS, is composed of three slabs, the two outer are 
alike and are cut at the ends, R and S, at an inclination of 45 ° to 
the length and to the incident light. 

Pm& 3^5 



Thus the beam, QRST, is totally reflected at R and S, by these 
two end faces. Between these slabs lies a central one of the 
same composition and thickness cut at R at 45 like the others, 
but cut square at the other end, and projecting as the diagram 

Fig. 242 

indicates. Thus the central portion of the beam, QR, is totally 
reflected at R, and transmitted out at the other end, and is ab- 
sorbed by the blackened walls of the box. The beam, Q X S, is 
totally reflected at the end, S, by.the upper and lower slabs, 
and the light absorbed by the walls of the box, but the central 
portion is transmitted in the direction, ST, by the central slab. 

Thus a tripartite field is produced and may be observed from 
the end of the tube, F ; the outer portions are illuminated by 
light which has entered at Q, and the central portion by light 
from Q 1 . 

The instrument is constructed and the prism and Nicol, N 1 , 
chosen to cause the light within the box to suffer approximately 
the same absorption along the two paths, QRS and Q X S. 

The tube, F, carries a second Nicol which is not shown, which 
acts as an analyser and which can be rotated by means of the 
divided metal circle, G. This circle carries two scales, one 
marked in degrees and the other giving ' densities ,' a term which 
will be explained below. 

When the instrument has no zero error the degree scale reads 
zero when both Nicols are parallel. In this case the light already 
polarized by N x is transmitted by N 2 , while the unpolarized 
outer portions of the field are polarized by N 2 in the same way 
as the central portion by N r 

Thus when there is no absorbing medium between the source 
and one of the windows, the field appears under these circum- 
stances of the same intensity in the outer and inner portions. 

F also carries a condensing lens which may be adjusted by the 
rod, R, projecting downwards from the tube. 

The purpose of this lens is to converge the light so that on exit 
all the light may enter the pupil of a normal eye. This is an 
important condition with which accurate photometric apparatus 
must comply in order that intensity comparisons may be of any use. 


The rod carries a scale past an index mark in the slot, B, on 
which are engraved two sets of numbers which have to be as- 
sociated in pairs. The upper scale records the distance of the 
source from the front of A in cms., and the number below this 
record on the lower scale denotes the maximum breadth of the 
source which is permissible if the photometric condition is to be 

Beyond the circle, G, the tube carries a lens system which 
focusses the light on to the slit of the spectrometer, and if it is 
desired to examine the field of view directly by eye an additional 
eyepiece is fitted into the tube. 

The apparatus requires careful adjustment which may be 
carried out as follows : 

Remove the prism from the spectrometer, illuminate the colli- 
mator slit and place on the prism table a piece of plane mirror or 
a right-angled totally reflecting prism, and adjust it until an image 
of the slit lies on the cross-wire of the telescope. Now place the 
source to be used with the photometer at the distance from the 
collimator slit at which it is to be situated during the experiment, 
having regard to the photometric condition which limits the 
distance to some extent on account of the size of the source. 
Adjust the source until an image of it lies in the centre of the 
field of view of the telescope, the collimator slit being now wide 
open and the eyepiece removed from the telescope, so that the 
source appears to lie at the centre of the object glass. Slide 
the photometer into position with the end of the tube, F, as 
nearly as possible 1*4 cms. from the slit, and with the window, 
Q 1 , directly between the collimator slit and source. Cover up 
the window, Q. 

Adjust the photometer by means of the three screws on which 
it stands, and by rotation about a vertical axis until the image 
of the source lies once more in the centre of the objective of the 

Place the plate, C, into position at a distance of 19 cms. from 
the source. The distance between P and P 1 is 3-8 cms., so that 
by moving the source a distance 1*9 cms., it can be brought to 
lie opposite the middle point of PP 1 and it should then be in the 
correct position. - 

This may be judged first by observing if bright circular patches 
of light lie symmetrically round the windows, Q and Q 1 . 

Make sure that, with the rod, R, adjusted so that the index 
lies opposite the mark denoting the distance between the front 
of A and the source, the width denoted by the lower reading is 
greater than that of the source. If this is not the case the source 
has either to be displaced* farther from A or diminished in size. 
Place the constant deviation prism and eyepiece of the telescope 



in position and adjust the spectrometer correctly for sodium 
light as indicated on p. 343. In doing this a strongly coloured 
Bunsen flame may be placed just closer to C than the light source. 

Make the line as sharp as possible by rotating slightly the 
milled head at the end of F. 

Now remove the Bunsen flame and open the slit to let in a 
convenient quantity of light. It will be probable that the 

Fig. 243 

central part of the field of view is slightly displaced with respect 
to the outer parts. This is exaggerated in the upper part of 
fig. 243. This may be corrected by a further small rotation of 
the photometer about a vertical axis. The three parts of the 
field should be separated by fine dark lines. If these are too 
wide they may be made narrow by adjusting the base screws of 
the apparatus. 

Correction of Zero Error 

In practice it is usually found that when the apparatus is set 
at zero the three parts of the field are not uniformly bright, and 
that the error is not the same for different wave lengths. In 
order to correct for this the readings of the apparatus are recorded 
when the field is uniform for different wave lengths. A shutter 
eyepiece is fixed to the telescope to cut down the light except 
over a narrow central strip, the wave length for which is recorded 
on the drum. When the central portion is the brighter, rotation 
of the analyser cuts it down, and the readings on both sides of 
the zero are observed and the mean taken for a series of different 
wave lengths and a curve plotted, showing the relation between 
error and wave length. 


These readings are observed on the degree scale or density 
scale as may be required. 

If the central part of the field is the darker it is not possible 
by merely turning the analyser to bring about uniformity. In 
this case a weak absorber, e.g. a plate of glass, is put into the path 
of the beam just before it enters Q. The thickness is chosen 
so that in the zero position the central field is just stronger, when 
the correction may be made as before. This should not often 
occur because the instrument is made so that the central portion 
is brighter than the outer, so that as the apparatus deteriorates 
with time the central portion may still be the brighter and the 
interposition of the weak absorber may be avoided. 

The substance of which the absorption is required is placed 
in the path of the light before it enters Q. 

In the case of a liquid it is necessary first to measure the 
absorption of the vessel containing it. 

The observations consist of noting the readings on the scale, G, 
when the field is uniformly bright for both directions of rotation 
of G. 

Let I denote the intensity of the light entering at Q 1 , and 
let I denote the intensity of that which enters at Q. The density 

scale records the values log ~, i.e. the logarithm of the ratio of 

the intensity of the light entering the medium to that transmitted 
by it. 

If this number be divided by the thickness of the material 
traversed a quantity known as the ' extinction coefficient ' is 

Let a denote the amplitude of the polarized light transmitted 
by Ni and suppose that the analyser is turned through an angle, B, 
from the position in which it is parallel to N r The amplitude 
of the vibrations transmitted by the analyser is therefore a cos 0. 

When the field is uniformly matched the intensity is I, and 

Io_ a 2 . 
I a 2 cos 2 ' 

.-. log ^-= log sec 2 6. 

This shows the relation between the density scale and scale 
of degrees. 

A solution which gives a characteristic absorption curve is 
one of eosin in alcohol. Eosin may be obtained from a bacterio- 
logical laboratory, and the solution must be very dilute, or so 
much absorption takes place that the transmitted light is very 


Carry out the determination described above and plot on the 
same curve the values of the ' density,' log —, and the cor- 
responding wave lengths, for the zero error, for the vessel alone 
and for the vessel and solution. 

From these it is possible to obtain the value, log^ for the 

solution alone. 

Fig. 244 shows the result of an experiment with eosin. Some- 
times the variations occur very rapidly and a curve is obtained 
like that drawn in diagram 243. Unless frequent observations 
are made in the neighbourhood of AB the variations at C and D 
may be missed. Whenever the curve shows any sign of change 
the neighbourhood where this occurs must be examined with 

AlosoY^Hon Curue for a u>cak 
Solution of Eostn >n AlcoKo\. 

the. o|p|per cwtye. \s ■&><• 
tElo&n -v- Glass etc. 
"The \oujer curve. i» •¥<* 
/toro covrecC\or\ <wA cjVass 
cxbsor^Hon . 

■4400 *=\BOO 5EOO B600 6000 

WaveWr^Yhe x \o cms 
Fig. 244 

It may not be necessary to determine exactly the value of 

log y for the solution, and in this case it is sufficient to make 

the zero correction and that for the containing vessel together, 
the whole appearing as a combined correction curve. 


To Find the Frequency of a Note by means of the Siren 

In this instrument a musical note is produced by puffs of air 
following one another in rapid and regular succession. The 
series of puffs is produced by blowing air through a number of 
holes in a rapidly rotating plate. 

The diagram (fig. 245) illustrates the instrument. It consists 
of a cylindrical metal chamber provided with a tap through 
which air can be blown. In the upper end is cut a series of holes 
lying regularly spaced on a circle with its centre on the axis of the 

Fig. 245 

cylinder. Above this lies a circular metal disc provided with a 
similar series of holes which fit above the former. The disc is 
mounted so that it can rotate about the axis of the cylinder 
and so alternately cover and expose the lower series of holes. 


SOUND 393 

The two series of holes slant in opposite directions as the figure 
shows, and when a current of air is blown into the chamber and 
the disc given a slight rotation, the puffs of air on escaping produce 
a pressure which drives the disc. 

By adjusting the influx of air the regularity and speed can be 
controlled so that notes of varying frequencies can be produced. 
If N is the number of revolutions made by the disc per second, 
and the number of holes is n, the frequency is N». In order to 
measure the number of revolutions the disc is provided with a 
metal bar, provided with a screw at one end, which works two 
dials, one registering units and tens, and the other hundreds of 

This form of apparatus is due to Cagniard de la Tour ; but it 
has the disadvantage that the speed can only be increased with 
greater air pressure and a consequently louder note. It is also 
difficult to keep the speed uniform. 

It is preferable to drive the disc with an electric motor, of 
which the speed may be regulated by including a resistance in 
the supply circuit, and the holes should be cut normally to the 
disc in order to avoid air pressure in the direction of the rotation. 

The siren gives a large number of harmonics, and it is necessary 
carefully to single out the fundamental note. 

Let it be required to find the pitch of a given note, as, for 
example, that produced by an open organ pipe. 

Carefully adjust the speed of the siren until beats are heard 
between the note it gives and that of the organ pipe by blowing 
at a particular pressure from a bellows connected to the chamber. 
The blower should then endeavour by a slight change of pressure 
to produce from the siren a note giving no beats. To some 
extent the frequency may be controlled by the tap, but it is 
important to keep a steady pressure on the bellows. 

At the same time a second observer should measure the speed 
of revolution by observing the number of revolutions recorded 
on the dials in a definite time (30 or 60 seconds). 

Verify the result of the determination of frequency by measur- 
ing the length of the pipe and its diameter. For an open organ 
pipe emitting the fundamental the wave length is approximately 
twice the length. 

The correction necessary to obtain a more accurate result 
is to add «6 radius for the open end and 2*8 radius for the flute 
mouthpiece. If the length of the pipe be I, and the radius of 
the pipe, r, the half wave length is given by : 

- = I + 3'4^ 
or X = 2(1 +■$•#). 


The velocity of sound in air for ordinary temperature may be 
taken as V = 33,300 cms. per second, or make the accurate 
correction for temperature by formula (A), p. 416. 


Thus the frequency is -•• 

Determine the frequencies of several organ pipes theoretically 
and experimentally, and draw up a table recording the speed of 
rotation of the siren disc, and the observed and calculated 
frequencies in each case. 

The Tonometer 

In Scheibler's tonometer a number of tuning forks are arranged 
in ascending order of frequency, each of which gives the same 
number of beats with its neighbour. The forks thus form a 
series in which the frequency increases by equal steps, and they 
are arranged so that the highest frequency is twice that of the 

In Appunn's tonometer the forks are replaced by reeds set in 
vibration by a blast of air from bellows of large capacity, and the 
apparatus has the appearance of a small harmonium provided 
with a series of stops by means of which any note may be 

This form of apparatus is not so accurate as the original one, 
for Lord Rayleigh has shown that the frequency of a vibrating 
reed is to some extent affected by the vibrations of its neighbours. 
As it is necessary to vibrate two successive notes in the experiment 
we have no longer a constant register of frequency as in Scheibler's 

It is first necessary to find the absolute frequency for each note 
on the instrument. Suppose there are (k + 1) notes, and con- 
sequently k intervals between them, and that the frequencies 

are N x , N 2 , . . , N* +1 , 

beginning from the lowest. 

If the number of beats be observed between all the successive 
notes, and be denoted by 

n x , w 2 , . . . n k , 

respectively, we have the following relations : 

N a -N 1 =» 1 , 
N 3 — N 2 = n 2 , 

N* +1 — N* = n k . 
Then adding both sides : 

N* +x -N 1 = # 1 + » l + ...+» 4 . 

SOUND 395 

But N* +1 = 2N X ; 

/. Nj = n x + n 2 +. . . + n k . 

We thus find N x by counting the successive numbers of beats, 
and we can then deduce from the equation the frequencies of 
all the other notes of the series. In practice, of course, the 
numbers of the beats will vary between the successive notes in 
different parts of the scale to a slight extent. 

In order to determine the frequency of the lowest note, count 
the number of beats at five or six different parts of the range and 
deduce the average difference of frequency. 

Let this be denoted by n. Then the frequency of the lowest 
note is k x n. 

After this determination has been made the frequency of any 
given note coming within the range may be determined. For 
example, suppose the frequency of a fork is required. Find by 
trial the note nearest to it in pitch and count the number of 
beats when the notes are sounded together. 

If this number be denoted by x, and N/ is the frequency of the 
note nearest the unknown, the frequency of the latter is N, ± x. 

The frequency of the note next above N, is N, + n, and on 
sounding this and the unknown note together the frequency of 
the beats will be less than n if its pitch is higher than that of N/, 
and greater than n if its pitch is lower. This enables a distinction 
to be made between ± x. 

In making this last part of the experiment it is advisable not to 
rely on the accuracy of the average number of beats for each 
interval, but to measure separately the frequency of the beats 6i 
the notes immediately above and below N/. 

It will then be easy to decide on the exact position of the 
unknown pitch above or below N/. 

The Determination of the Frequency of a Tuning Fork by the 
Method of the Falling Plate 

A smoked glass plate, P, is suspended vertically by a piece of 
thin string or thread over two nails, QQ, the thread being attached 
to the upper edge of the plate by means of sealing wax or by any 
other convenient method. 

The fork is held in a clamp, H, and carries a light style of 
bristle or thin aluminium wire attached to one prong by as little 
wax as possible. The style is just in contact with the plate so 
that when the plate falls it removes some of the soot and leaves 
a trace. 

In order to prevent breakage a padded wooden stand, AB, is 
placed just below the plate. 
The fork is stroked gently by means of a violin bow and the 



plate is allowed to fall straight down by burning the thread 
between the nails, QQ. 

A wavy line is traced by the style similar to that shown in 
fig. 247, but with more waves, and usually of small amplitude. 

A point, O, is chosen just clear of the indistinct portion drawn 
when the plate was moving down in the first stage of the motion, 
and consequently before its velocity had sufficiently increased 

Fig. 247 

to open out the waves, and from it is counted a number, n, of 
complete waves to the point S. Again, n waves are counted 
to the point S r Let the spaces, OS and SS X be of lengths, s and s lt l 
respectively. Then the time taken to fall over these two lengths 
is the same, let us say t. 

If N denotes the frequency of the vibrations : 

n = N*. 

SOUND 397 

Let u denote the velocity of the plate at the instant correspond- 
ing to the mark, O. 
Then by the equation for space described under acceleration, g, 

s = ut + \gP, 
and s -f- s x = zut + 2gt 2 , 

for the time of description of s + s t is 2t. 
Hence Sl — s = gt 2 , 



N = n • J- g 

vs x — s 

"1 " 

The distances, s and s v are measured carefully by means of a 
travelling microscope, and the value of N obtained is that for the 
fork vibrating with the load consisting of the style and wax and 
affected by the friction of the style against the plate. 

This should be allowed for by taking a second fork of nearly 
the same frequency as that under examination, before attaching 
the style, but of slightly higher pitch. Carefully load this fork 
by adding wax until no beats are heard when both sound together. 

Then when the first fork is loaded and has the style touching 
the plate as in the experiment, again sound the two together 
and count the beats. The number of these per second gives the 
number of periods lost per second on account of the loading and 

This number added to the value of N, determined in the 
experiment, gives the corrected frequency. 

In marking off the points O, S, and S v be careful to choose 
them at corresponding points of the waves. 

If O is at the summit of a crest, S and S x must lie in a similar 
position n and 2w waves later, respectively. 

The plate may be conveniently blacked by holding it just over 
a turpentine flame which gives a good deposit of soot ; the 
flame of a paraffin lamp gives also a satisfactory deposit. 

Chronographie Methods of Determining the Frequency of a Fork 

In both the methods to be described under this title a fork is 
set in vibration and, electrically maintained, with a style attached 
to one of the prongs lightly touching the blackened surface of a 
cylinder which can be rotated about its axis. 

On rotating the cylinder the track made by the style appears 
as a wavy line which can be opened out so that each vibration is 
distinctly separated from its neighbours by adjusting the speed 
of rotation. 


The drum may be rotated by the handle, H, or by a falling 
weight not shown in the diagram, which rotates the drum at a 
convenient speed. 

In this case the rotation is regulated by a governor, and by 
releasing a catch, the handle, H, can be employed for rapid 
movement of the drum. This also is not shown in the figure. 
The drum moves along an axis cut with a screw thread, so that 
the track drawn by the style forms a wavy helix round it. 

Fig. 248 

It is then necessary to have some record of time with which to 
compare the vibrations. The two methods differ in the way 
the time is obtained. 

In the first a stand is mounted conveniently near to the fork, 
as illustrated in fig. 248, which carries a pointer actuated by an 
electromagnet, M. The pointer, P, is carried on a lever one end 
of which consists of a strip of iron or steel, which is attracted by 
the electromagnet core when a current flows through the exciting 
coils. When the current is off the lever is held in a position with 
the strip a short distance from the magnet by means of a spring. 

Thus, as the cylinder rotates, a line is drawn round the surface. 
On passing the current in the coils the pointer moves and makes 
a kink in the line. If the current is put on at regular and known 
intervals, a time record along the side of the waves is produced 
on the drum. Thus by counting the number of waves between 
consecutive kinks the frequency of the fork may be deduced. 

The regular intervals are obtained by connecting the wires, 
AA, to a battery, and completing the circuit by means of the 
mercury cup, M, and the pendulum, P, as shown in fig. 249. 

Mercury is poured into a hole cut in wax, W, so that it stands 
just above the wax surface, and the wax so placed that a strip 
of wire hanging down from the pendulum just touches the surface 
as it passes its lowest position. 

The circuit is completed twice in each complete period, and at 
each instant the pointer makes a record on the drum. 

It is improbable that the interval between successive records 
will be one half a period, since this would require exact coincidence 
of the point of contact with the mercury and the lowest point of 



the swing. The time between alternate records will, however, 
measure the time of a complete period. 

Thus, in counting up the vibrations, find the mean number 
between alternate records. 

Obtain a long helix, begin at the first stroke of the pointer, 
and count the number of vibrations up to some later odd numbered 
stroke. Find the mean number per complete period. Repeat 
this, beginning with the second stroke and ending at some later 
even numbered stroke, and find the mean number again. The 
two values should agree, but if there is a slight variation take 
the average value of the two results. . 

Fig. 249 

To obtain a blackened surface, take *a sheet of smooth white 
paper, and wrap it round the drum, one layer thick, holding it 
in position by means of gummed paper. To blacken it, rotate 
it over a turpentine flame, or coat it with camphor smoke. 

The coating of soot should not be very thick, the style will 
then remove the soot and leave a white wavy trace. 

The paper may be smoked again when once used and the track 
covered up. 

The time of a complete period is measured in the usual way, 
by timing the pendulum. 

The frequency obtained is, of course, subject to a correction 
similar to that of the last experiment, and this should be 
determined and applied in the manner described. 

Fig. 250 represents a convenient method of carrying out the 
determination in an alternative way. 

The cylinder is held vertically, and is rotated by a handle or 
string round a drum as shown. 

The fork is maintained electrically as before, but the axle of 
the drum is connected to one secondary terminal of a small 


induction coil, while the second terminal is connected to the fork. 
The primary circuit is completed through a pendulum and 
mercury cup as before. Each time the primary circuit is com- 
pleted the induction coil is excited and a spark passes to the drum 
from the style, knocking off a little soot and leaving a white dot 
to record the instant of closing of the primary circuit. The 
same procedure and precautions are adopted as before. A thin, 
flexible copper wire will be found suitable for suspension of the 
pendulum, and with a small induction coil there is no incon- 
venience on account of shocks obtained when the apparatus is 

Fig. 250 

This method was employed by A. M. Meyer. It should be 
noted that uniformity of motion of the drum is not necessary. 
The methods are both inferior to those in which an optical method 
is employed for determining frequency as in Rayleigh's method. 

The Frequency of a Tuning Fork by the Stroboscopie Method 

The fork is fitted with two very light plates fastened at the 
extremity of the prongs, one to each, and so that one may vibrate 
freely past the other. These plates may be of thin cardboard or 
aluminium, so that the loading affects the fork to the smallest 
possible extent, and they are stuck on to the fork by means of a 
little wax. A disc is taken provided with a number of dots 
placed at equal intervals round circles concentric with the disc. 
Each circle has its own interval length. 

SOUND 401 

The fork is maintained electrically (p. 138), and the disc 
placed behind it with its dotted surface brightly illuminated. 
Each plate attached to the fork is provided with a slot, and 
when the fork is at rest the slots lie directly behind one another. 
Thus the disc can be seen through them. 

The slots and one of the circles of dots are so placed that the 
dots can be seen by looking through the slot, as the disc slowly 
rotates. If the fork is vibrating it is possible to see through the 
slots twice in each complete period, and thus 2,n times per second, 
where n denotes the frequency. 

The disc is caused to rotate uniformly by means of an electric 
motor provided with a resistance in circuit to vary the speed. 

The speed is gradually increased by adjusting the resistance 
until when the fork is vibrating the dots in one of the circles 
appear to be at rest. 

In this case a dot moves up as the disc rotates, so that each 
time the slots are in line a dot is just in line with them. The 
eye sees apparently one stationary dot, and the effect of rotation 
is lost by looking through the slot. 

In order to count the number of rotations per second made by 
the disc, a counting arrangement is attached and the time taken 
over a definite number of rotations, when the dots remain 
apparently steady, by means of a stop-watch. 

Let there be N rotations per second in the case when the dots 
belong to a circle containing p of them. 

In this case the time taken by a dot to take the place of the 

one preceding it is : ~-r second. This is equal to one half the 
period of the fork. 

Hence — = — — 

N^> 2 n ' 

or n = - Np. 

Now further increase the speed of rotation until once more the 
dots appear steady. In this case the speed of the disc is such 
as to cause a dot to take the place of the dot two intervals in 

front. The time taken to do this is ^n, where N 1 is the 

N 1 ^ 

number of rotations per second 

From this we may calculate the value of n once more. 

On further increasing the speed until a dot takes the place of 
another three intervals in front we can obtain a third calculation. 
Repeat this for several series of dots. 

With the fork loaded it is necessary always to correct for the 
loss of pitch due to loading as before. 



We may, however, avoid this by making one prong of the fork 
bright over a small area, and by rotating a disc provided with 
several series of concentric holes in front of it. If the fork is 
well illuminated and the disc carefully mounted, it will happen 
that for some particular speeds the fork appears stationary when 
seen through the holes. 

The experimental details and the mode of deduction of the fre- 
quency of the fork from the observations are similar to the former. 

The observations in stroboscopic experiments can nowadays 
be made much more conveniently by the use of a neon lamp than 
by the method described above. 

This lamp has the property that it lights up immediately a 
voltage is supplied to it without any appreciable lag, and is 
extinguished immediately when the voltage is taken away. 

The lamp consists of a flat piece of aluminium with a rod of 
the same metal lying a short distance from it and parallel to it 
in a bulb containing neon at low pressure. 

When included in the secondary circuit of an induction coil, 
it lights up each time the current in the primary is broken. There 
is no effect at ' make ,' because in the construction of the usual 
type of induction coil the effect at make is suppressed and that 
at break of the primary circuit intensified. 

If the primary coil is put into the circuit which actuates the 
fork the primary current is made and broken once per vibration 
of the fork and the lamp flashes out once per complete vibration. 

Of course, the make and break attached to the primary must 
not be allowed to work, the fork takes its place. All that is 
necessary is to make the connexions in the usual way and to screw 
the platinum-iridium point close up to the clapper to prevent 
separation and to place some object, e.g. a small block of wood, 
between the soft iron on the clapper and the end of the armature 
to prevent oscillation as the current fluctuates. 

The lamp is used to illuminate the rotating disc provided with 
a series of dots on a white background and the speed adjusted 
until one row of dots appears stationary. 

When this is the case one dot just moves up to take the place 
of a dot somewhere in front of it during the interval of darkness 
between the flashes of the lamp, i.e. during the period of vibration 
of the fork. 

The calculation of frequency is made as before. 

It is a great advantage to be able to avoid loading or marking 
the fork and to have the disc in any convenient position where it 
may easily be observed. 

Instead of a series of concentric circles with regularly spaced 
dots, it is more convenient to draw on a circular disc a series of 
concentric regular polygons. A small triangle is drawn just 

SOUND 403 

about the centre of the disc, about this a square, then a pentagon, 
and so on. c The triangle may be coloured white, the space 
between it and the square blackened, the space between the 
square and pentagon coloured white, and so on alternately. 
When the/speed is adjusted exactly one of these figures appears 
stationary and the frequency is easily calculated. By varying 
the speed the figures may be made to appear stationary in turn 
and several determinations of frequency made. 

The Composition of Two Simple Harmonic Vibrations in the Same 
Direction (Beats) 

An apparatus which will combine graphically two simple 
harmonic vibrations in the same direction has been invented by 
Koenig. It consists of a large fork mounted on a stand and 
provided with a clamp by means of which a strip of glass can be 
held horizontally and fastened to one of the prongs (fig. 251). 


Fig. 251 

AB denotes the strip and C a weight, attached to the other 
prong for the purpose of balancing. The glass is coated with a 
thin layer of lamp-black, by means of a smoky flame. 

A second fork is mounted above the first and carries a light 
style adjusted so that it just touches the glass plate. 

This fork is fixed to a sliding base by means of which the style 
can be drawn along the smoked plate. 

If both style and plate are vibrating a curve can thus be 
traced which represents the motion of the upper fork relative to 
the lower. 

Thus if at any instant the lower fork is displaced a distance, y, 
from the standard position, and the upper is displaced a distance, 
y\ the displacement of the style over the plate is {y 1 — y). 
The forks are made to vibrate with nearly equal amplitudes. 
This may be done by bowing or by using a strip of metal which 
is wide enough to open out the forks to a convenient extent ; 
the metal is then quickly removed. The slider is drawn along, 
not too quickly, and the trace obtained examined. 

> When the forks have nearly the same frequency this will con- 
sist of a wavy line with waves of varying amplitude. In this 
case, with the amplitudes of the forks nearly equal, the smallest 
waves will have almost zero amplitude, and the fluctuation 
expresses graphically what the ear recognizes as beats. 


The time between consecutive minima measures the interval 
between the beats. 

If the glass plate is long enough three or four intervals extending 
between consecutive minima will be obtained. The speed of 
motion of the slider must be adjusted so that the individual 
vibrations are drawn out to an extent which enables them to be 
counted easily. 

It is best to count the number of intervals on the plate and the 
total number of vibrations between the first and last minimum 
points. This enables the mean number of vibrations between 
consecutive beats to be obtained. 

Let this number be x. 

We shall suppose that the two vibrations have frequencies n 
and (n + m) per second, so that they may be represented by : 

y 1 = a sin 2% (n + m) t 
and y = asm. (2nnt + a), 

where a is included to take account of any difference in phase 
that may exist when the vibrations begin. 

The resultant displacement recorded on the plate is : 

Y = (yi — y) = a {sin 2-n (n + m) t — sin (2nnt + a) } 

= 2a cos \2rtln -\ — \t +- isin (271— t — -V 

In the cases when beats are heard, m is much smaller than 
n or in + -\ so that the simplest way of regarding this expression 

is to consider it as a S.H.M. of amplitude : 

2a sin (2tc • \mt — |a) = A (say), 

and then Y = A cosj 2* \n + ~J t + £<x • 

represents a S.H.M. of amplitude, A, and frequency, \n + ~J. 

The amplitude, A, attains a maximum value, 2a, and sinks 
to zero alternately. It is zero for the values of t given by : 
imt — |a = o, re, 2tc, etc., 

i.e. for values of t = ^ — ^ + -. ^ + - , etc., 

i.e. at intervals of— sees., or m times per second. 
Thus the number of beats is m per second, the same number as 
the difference between the frequencies of the two notes. The 

frequency of the note heard is In + — V 

SOUND 405 

n + — J vibra- 
tions per second and m beats. The number of vibrations between 
two beats is : 

H» + -)• 

m\ 2 / 

But this is counted on the plate and found to be x. 

Thus — In -\ — ) = x, 

tn\ 2/ 

or n — m (x — \). 

The beats can be timed by means of a stop-watch. 

When the notes are sounding, count them for as long as possible, 
and take the time of the interval. If the number of beats counted 
is N and the stop-watch is started on the first and stopped at 
the Nth, the interval between the beats is : 


where T is the time interval recorded on the stop-watch. 

Thus -is known, and the above equation gives n. The 


frequencies of the notes are thus, n and (n + m). 

The method must be regarded as an illustration of the 
phenomenon of beats — it is not an accurate method for the 
determination of frequency. 




Fig. 252 

The Composition ol Two Simple Harmonic Vibrations Perpen- 
dicular to one another. (Lissajou's Figures) 

Let the co-ordinates of a point, P (fig. 252), be (x,y) and let P 
move so that 

x = a sin pt 

y = b sin (pH + a). 

The motion of P then consists of two simple harmonic motions 
along two perpendicular directions. 


When t is zero, x is also zero ; but y has the value b sin a. 
Thus the two S.H.M.'s are in different phases. 
From the equations it follows that : 

.1 . x 

t = — arc sin- 

p a 

and t = — arc sin \ — — • 

p 1 o p 1 

Hence between x and y there exists for all values of t the 
relation : 

i . x I . y a 

- arc sin - = -— arc sin \ — -:• 
/> a p 1 b p 1 

This represents the curve on which P lies. 

When the relation between p and p 1 is simple, as for example : 

p =p\ p = 2p\ 2p = 3P 1 , etc., 

the point, P, describes the curve completely in a short time, and 
afterwards retraces it. 
The case, pj= p\ is very simple, for we then have : 

x = a sin pt, 

y — b sin (pt + a). 

In the general case this represents an ellipse with its centre 
at the origin. 
If a = o, then : 

x y 
a = b' 

and P describes a straight line. 


If a = —, P describes the curve : 


a 2 ^ 6» 
This is an ellipse whose principal axes lie along OX and OY. 
In the case when the amplitudes of vibration are equal along 
both directions, a = b and the locus of P is the circle : 

x* + y 2 = a*. 

Thus when the periodic times are equal, in general P describes 
an ellipse ; special cases of this general case are the straight line 
and circle. All these curves are therefore appropriate to the 
case of equal frequencies. 

If OB is equal to b sin a, B represents the position of P at the 
time t = o, or at instants later by a complete period than this 
initial time. 



If the frequencies are not quite equal, let us say that the 
frequency along OY is a little the greater, then when another 
period is complete the displacement along OY will be a little 
different from that in the first case, i.e. P will lie at some point, B 1 . 

If OB 1 = b sin a 1 , 

a 1 is slightly different from a, 

and there has been a slight change in phase on the part of the OY 
motion. Thus a will continually vary, and will pass through all 
the values from o to 2n. This will cause a continual change in 
the shape of the curve described by P. It will sometimes be a 
straight line and sometimes an ellipse. If it happens that the 
two amplitudes are equal, we shall have a circle sometimes. 

The closer p 1 is to p, the slower will this change take place, 
so that by watching the movement of P we can test the closeness 
of the two frequencies. If the figure is maintained steady, 
without change, the frequencies are equal. 

Tfre same argument can be applied to the cases where one of 
the other simple relations exists between p and p 1 . 

When the shape of the curves corresponding to these relations 
is known the approximate ratio of the frequencies can be recog- 
nized and the exactness of the ratio tested by observing the 
rate of change of shape throughout the series. 

We can deduce the ratio of the frequencies by examining one 
of the curves, e.g. fig. 253. It cuts the Y axis in four points and 
the X axis in three, so that the frequencies are in the ratio, 4 : 3, 
for the vibrating point makes four vibrations parallel to OX in 
the same time that it makes three parallel to OY. 

Fig. 253 

This principle may be employed to investigate the way the 
period of vibration of a rod, fixed at one end, varies with the length 
of the rod. The apparatus consists of a vertical flat rod or spring 
with a lens fixed at its upper end and a horizontal spring carrying 
a screen with a small aperture (fig. 254). 


Light from a source, S, is focUssed by a lens, L 1? on to the 
aperture, A, while the lens, L 2 , carried by the vertical spring, Vr 
focusses an image of the aperture on the screen so that we have 
a bright point, P. 

Motion of the horizontal spring alone, causes P to trace a 
vertical line, and represents the S.H.M. of A. 

Motion of the vertical spring alone, produces a horizontal 
S.H.M. on the screen. When both springs vibrate together, the 
path of P represents the combination of the twb motions. 

The vertical spring is of fixed length, but the horizontal spring 
can be clamped at various points, so that the vibrating length 
can be adjusted. 

Make the first adjustment so that P describes an ellipse, straight 
line, or circle without change of form. 

In this case the periods are the same. 

Now change the length until another steady curve is obtained 
without change of form. Draw it carefully and deduce the 
ratio of the periods of vibration. 

Make several determinations of the ratio of frequencies and 
corresponding lengths, and draw a curve showing the relation 
between the length and frequency, taking the vertical spring as 
a standard. 

Fig. 255, shows the curves described by P for a few frequency 
ratios which will serve for reference. 

The Vibration Microscope 

The essential features of this apparatus are the same as those 
described in the last experiment. The two vibrating springs are 
replaced, one by a fork and the other by a vibrator, the frequency 
oi which is to be compared with that of the fork. 

The second vibrator is sometimes another fork or violin string. 



A bright source of light such as a speck of chalk is attached to 
the second vibrator, while the fork carries, attached to one prong, 
a lens which forms the object glass of a small microscope. 

If the lens alone vibrates, on looking through the eyepiece the 
motion of the chalk is simple harmonic, on account of the vibra- 





Fig. 255 

tions of the lens. If the other vibrator is in motion and the lens 
is at rest, the motion observed is, of course, that of the vibrator 
alone. These two motions are arranged to take place in two 
perpendicular directions so that a figure of the type described in 
the last experiment is observed. It is steady if the frequencies 
are exactly adjusted, but goes through the appropriate series if 
the frequencies are not identical. The rate of progress through 
the series may be observed and determined by means of a stop- 
watch. After one completion of the cycle there has been a gain 
of a whole vibration by one vibrator over the other. If the 
frequencies be N and N 1 , and the time for completion of the 
series is t seconds, then 

N~ N l = 


for N2 and N 1 ^ are the numbers of vibrations made respectively, 
and these differ by one. 

If the fork be slightly loaded we can find which is the greater 
of the frequencies by again observing the rate of progress through 
the cycle. If this time is shorter than before, the time of gaining 
a period is less than before and the frequency of the fork is the 

We shall describe how the apparatus may be used to find out 
the character of the vibrations performed by a stretched string 
in the manner in which Helmholtz used the apparatus to examine 
the vibrations of a violin string. 

P L P 


Fig. 256 


The string, SS, is stretched below the prongs of the fork, PP, 
to one of which the lens, L, is attached. The part of the string 
to be examined is slightly blacked by ink, rubbed with wax, when 
dry, and powdered with starch or chalk. A few white particles 
will remain sticking to the string, and one of them is illuminated 
by a lamp and focussed by the microscope and its movements 
observed. The tension of the string is adjusted until the figure 
apparently described by the chalk, as seen in the microscope, 
remains steady. The vibrations of the fork are electrically 
maintained while the string may be bowed or plucked. 

The frequency of the fork is known, so that, as its motion is 
simple harmonic, we can find the displacement due to it at any 
of the instants during the vibration. The string vibrates at 
right angles to the direction of vibration of the fork, so that 

Fig. 257 

from the curve obtained we can subtract the vibrations of the 
lens and draw a curve showing the displacement of the string 
at different times. 

The white speck is first obtained in the centre of the field, and 
its mean position represented by the origin, O (fig. 257). 



The curve is drawn to a convenient scale accurately from 
measurements observed by the scale in the eyepiece. If the 
microscope is not furnished with a scale, throw an image of an 
illuminated scale to lie coincident with the string and view the 
white speck and image together. 

In the figure, OA denotes the amplitude of vibration of the 
fork, its motion being assumed to take place along AOA 1 . 

If its frequency is n, the motion is given by : 

x — OA sin 2nnt. 

Thus for any time, t, we can find the displacement, ON, along 
the x-aods. At this instant the displacement along the other 
direction, i.e. due to the motion of the string, is NL. Thus by 
observing several values of the ordinates and the times corres- 
ponding, we can plot a second curve showing the time-displace- 
ment for the string. Its shape will indicate the character or 
quality of the note emitted. 

Fig. 258 shows the observed curve and the time-displacement 
diagram for a stretched string when bowed. The first curve was 
observed while the string was being bowed. The bow is drawn 
slowly and regularly across the string. Slight fluctuations are 
liable to occur during this process, but they appear as slight 
variations of a figure remaining on the whole permanent. The 
dimensions of this figure were obtained. 

Fig. 258 

The deduction of the diagram for the string is made in the 
following way : 

Draw on squared paper the figure observed (fig. 258), and 
draw the extreme vertical tangents, AP and A 1 ? 1 . We are 
assuming that the vibrations of the fork are executed along AA 1 . 

With the middle point of AA 1 as origin, describe a circle on 
AA 1 as diameter, and divide the circumference into a convenient 
number of equal parts. In the diagram the number is twelve. 

If a perpendicular be drawn from any point on the circum- 
ference of this circle on to AA 1 , the displacement of the foot of 
this perpendicular from O will represent the displacement of the 
prong of the fork from its mean position. 

Along the line, A*A, produced, beginning at o, twelve equal 


intervals are marked off, as shown, to represent intervals of time 
corresponding to the points on the circle. Beginning at A, draw 
ordinates to the curve passing through the points marked on the 
circle. For convenience in drawing only one of these, that 
through the point, 8, is drawn. Let this cut the curve at the 
point, Q, and the line, AA 1 , in N. Then NQ denotes the dis- 
placement of the string from its central position in magnitude 
and direction. 

This displacement is plotted at the point, 8, on AA 1 produced, 
and we thus obtain a point, q, on the displacement diagram for 
the string. This process is carried out for all the points on the 
circle, and the diagram plotted. 

It will be noted that the ordinate, NQ, cuts the curve in a 
second point, B. There is no doubt as to which point is to be 
taken when actually drawing the curve, for we begin at A and 
pass along one branch of the curve and back along the other. 

In this discussion we have associated the upper part of the 
curve from P to B and then to P 1 , with the times corresponding 
to the points from o to 6. 

Transverse Vibrations of Strings. (Melde's Experiment) 

The object of the experiment is to verify the laws of vibration 
of a string under tension. In such a case a disturbance travels 
along the string with a velocity, v, given by : 

\ m 

T denotes the tension expressed in absolute units, i.e. poundals 
or dynes, and m is the mass per unit length. 

When the string is fixed at both ends, there is a node at each 
end in its fundamental mode of vibration, with a loop or antinode 
in the middle. The corresponding wave length is twice that of 
the string. If this wave length is denoted by \ Xt we have : 

ni== t \S' 

«! denoting the frequency of this note. 

This mode is illustrated in fig. 259. The string may also vibrate 
to produce the overtones or harmonics as illustrated in figs. 260 
and 261. 

In these cases, if the frequencies are n 2 and n z , and the cor- 
responding wave lengths x 2 and x 3 , we have : 




x 2 \w' 

and Wo = 

and if / denote the length of the string : 


n x : n 2 : n s : 

Fig. 259 

In the experiment the string is set in resonant vibration by 
impulses having the same frequency as one of its modes of 
vibration. For this purpose ordinary string is unsuitable. It 
is not uniform and does not divide into equal segments; but it 
will be found that a length of fishing line is satisfactory, as a 
rule it is sufficiently Uniform. 

One end of the cord is attached to the prong of a tuning fork, 
by tying it to a small wire hook soldered on to the prong, or to a 
small screw which is held in a hole bored in the prong. 

The other end passes over a small pulley and carries a weight 
which produces and measures the tension. 

The vibrations of the fork are electrically maintained (see p. 138) 
and by properly adjusting the length and tension the string can 

be made to break up into stationary undulation with well-defined 

The fork may be placed so that the motion of the prongs is in 
the direction perpendicular to the string (fig. 262), or along it 
(fig. 263). 


In the former case the frequency is the same as that of the 
fork, in the latter it is half as great. 

For when in the second case the prong is in the extreme position 
on the left the string is slack in the first vibration, and when in 

Fig. 263 

the extreme position on the right it is horizontal and tight. The 
inertia of the string carries it onward so that when the prong 
returns to the extreme left position, and thus completes one 
vibration, the string completes a half vibration. 

The student should examine the interesting effects produced 
when the prong moves in a direction between these two and 
thus produces in the string a combination of the two modes of 

We shall consider the former case only in this description of 
the experiment. 

The frequency of the fork is denoted by N, and when the string 
is in resonant vibration, this is also the frequency of the mode of 
vibration of the string, corresponding, let us say, to a wave 
length, x. 

Then : N = £ J.2, 

x \ m 

£ =_!__ constant. 

Thus by varying the tension and consequently the wave 

X 2 
length, we should find ~ constant. 

To determine X, measure the distance between the first well- 
defined node on the right of the string and the last on the left. 
Let this be d and suppose there are k loops between, then, 

_ 2 d 

X_ k ' 

Note that the ends of the string at the fork and the pulley 

should not be taken. There is a certain amount of movement 

at these two points. 

Draw up a table showing values of X, T, and „• 

SOUND 415 


(a) The Determination of the Velocity of Longitudinal Waves along 

Kundt's apparatus consists of a glass tube about a metre long, 
and of diameter about 3 cms., provided with an adjustable 
piston near one end. The tube is supported horizontally on a 
table by resting it on two wooden V-shaped stands. 

Near the other end of the tube is a second piston, Q, attached 
to the end of a metal rod, DQ (fig. 264). This rod is clamped at 
its middle point, C. 

A _B c D 

— * 3TFT I"1 r~7 IS~P z 

Fig. 264 

For the purpose of the experiment the tube must be quite dry, 
and a light powder, such as lycopodium powder, is placed in a 
line at the bottom of the tuj>e extending along its length between 
the two pistons. A convenient way of inserting the powder, is 
to spread it along a metre rule, place the rule in the tube and turn 
it upside down. 

If the tube is not dry it must be warmed above a Bunsen flame, 
and a current of air blown through it. 

The metal bar, which may be of brass with a diameter of about 
•5 cm., can be set into longitudinal vibration by stroking it along 
CD with a piece of wash-leather and powdered resin. 

In the fundamental mode of vibration the ends, D and Q, are 
antinodes, and the fixed point, C, is a node. The wave length 
is twice the length of the rod. 

x = 2l. 
If the frequency of the note is n, and the velocity of the waves, v, 

n\ = v. 
If the distance between the pistons is L, the air between them 
will have a fundamental wave length, 2L, and overtones with 
corresponding wave lengths: 

L, f L, I L, etc 
The corresponding frequencies are : 

v y v_ 

2L' L' |L ' etC " 
where V denotes the velocity of sound in air. 

If one of these frequencies is the same as that of the rod, the 
air will be set into strong resonant vibration, and will move the 
light powder. This will settle down at and near the nodes where 


the air is least in motion. As many as possible of these should 
be used to find the average distance between two nodes. Choose 
as carefully as possible the position of a node at one end of the 
tube and locate the node nearest the other end. Measure the 
length between these two points, and divide by the number of 
spaces, such as NM (fig. 265). 


N M 

Fig. 265 

The^pattern will be somewhat similar to that in this figure, 
and the longest line of each set marks the nodal position. 

It will probably happen that there is not strong vibration of 
the air at first, but by slowly moving the piston, P, forward or 
backward, the length between the pistons may be adjusted so 
that resonance occurs. 

Twice the distance between the nodes is the wave length of 
the sound in air. The velocity of sound at o° in air is 33,060 cms. 
per second, and at a temperature, t : 

v, = 33060(1 +^y (^ 

Thus the frequency, n x — — ', where x 1 is the wave length in 


air. Since resonance occurs, 

n = n 1 ; 

1 _ Y« 

**• 2/ ~~ x 1 ' 

or v =^V t - 

In carrying out the experiment it is a good plan for one 
observer to continue stroking the rod, while the other carefully 
adjusts the piston until the powder moves violently and settles 
down into the pattern of fig. 265. 

(b) The Veloeity of Torsional Vibrations in a Rod 

If instead of stroking the rod longitudinally with the resined 
cloth, it is held near the end, D, and the cloth turned so that it 
slips over the surface in a direction that would cause the rod to 
rotate round its axis, a note is emitted of different frequency 
from that given when the rod is in longitudinal vibration. This 
note corresponds to torsional vibrations and will set the air in 
resonant motion as before. 

SOUND 417 

Find in this way the velocity of these waves. 

It is not easy to obtain a loud note by this method — the force 
applied should not be great, but, with a little practice, it should 
be possible to produce the note. 

(c) The Determination of Young's Modulus and the Modulus of Rigidity 

These constants may be determined from a knowledge of the 
velocity of longitudinal and torsional waves in the bar. 
The formula for the former is : 

and for the latter : 



where E is Young's Modulus, n the modulus of rigidity, and p 
the density of the material of which the bar is made. Its value 
may be taken from a table of physical constants. When ex- 
pressed in C.G.S. units, E and n are expressed in dynes per sq. cm. 
Find the values of E and n from the determination of the 
velocities in the previous experiments. 

{£) The Velocity of Sound in Carbonie Acid Gas 

Kundt's tube may also be used to determine the velocity of 
sound in gases. Suppose the tube to be filled with a gas in 
which the velocity is V and the wave length, \. 

Then the frequency is 

Vj V. 

\ X 1 ' 
• Yi _ h. — distance between nodes in the gas 
V ~~ X 1 distance between nodes in air. 

The procedure is, therefore, first to obtain resonance between 
the rod and air column, and to find the mean distance between 
the nodes, then to drive out the air and fill the tube with the gas, 
and again obtain resonance, and measure the distance between 
the nodes in this case. 

\ a is then found from the last equation. 

The gas must be quite dry or the powder will stick to the glass 
and fail to respond to the motion of the gas when resounding. 
It may be necessary to pass it through drying tubes before filling 
the tube. 

A slight modification of the apparatus is necessary for this 
purpose (fig. 266). 


In the former case the adjustable piston may fit loosely, but 
for the present purpose it must be both adjustable and gas-tight. 

It may consist of a cork round the outside of which a rubber 
band or piece of cloth is stretched. 



Fig. 266 

The cork carries a tube to admit the gas and the piston may be 
adjusted by means of this tube. 

The metal rod passes through a tightly fitting cork, also pro- 
vided with a tube which can be opened and closed by means of a 

We shall suppose that the velocity of sound in carbon dioxide 
is to be measured. 

Connect the source of gas to the inlet tube, I, and open the exit 
tube, E, to the air. Allow the gas to flow in steadily so that it 
will fall to the bottom of the tube and the air will flow over at E, 
also, so as not to disturb too much the powder which we assume 
has been spread out along the bottom of the tube. 

The experiment should be performed close to an open window 
or in a draught cupboard to prevent escape of the gas into the 

Continue the passage of gas long enough to ensure that the air 
is driven out and then close the inlet and outlet tubes by means 
of stop-cocks, and proceed as in the last experiment. 

Care must be taken that the temperature of the gas is the 
same as that of the air, unless the temperature of the gas is 
measured in some way. If the gas is delivered from a cylinder 
it will be colder than the air and it must be allowed to acquire 
the air temperature before closing the stop-cocks. Otherwise 
we shall be measuring the velocity in the air at one temperature, 
and that in the gas at another. Reduce the velocity to that 
at zero. 

To do this, note that : ^ = ^ = ratio of nodal distances, 

where the affixes t and o denote temperatures. 

In order to determine when the air is all driven out from the 
tube, collect a little of the gas issuing from E, in an inverted glass 
cylinder over mercury, and introduce on to top of the mercury 
column a little of a solution of caustic soda or potash. By 
noting how much of the gas is absorbed it can be seen if any air 
is left. The C0 8 is all absorbed by the solution and no gas should 
be left. 



(e) To Calculate the Ratio of the Specific Heats of a Gas 

From the result of the last experiment we may determine the 
constant, r, for carbon dioxide, i.e. the ratio of the specific 
heat at constant pressure to that at constant volume. 

For the velocity of sound in a gas at o°, is given by the formula : 


where p is the pressure and p the density of the gas at o° C. 
The value of p may be determined by the barometer since the 
tube has been filled at atmospheric pressure. 

p must be expressed in dynes per sq. cm. To make the cal- 
culation, take the density of mercury as 13*60 gm. per c.c, and 
P = '001974 gm. per c.c. 

The measurement of the distances between nodes may be 
performed simply by the ordinary use of a metre scale ; but a 
slight addition to the apparatus will add to the accuracy. 

A metre rule is fixed just below the tube and parallel to it, 
and sliding over the rule or along one of its edges is a wooden 
base, B, carrying a metal disc, D, with a hole, H, at the centre, 
and a frame, F, with cross-wires, C (fig. 267). 

Fig. 267 


H serves as an eyepiece and HC is aligned on the nodes marked 
out by the powder. 

An index on the base indicates the position of the stand. 

If the apparatus is aligned consecutively on the nodes, and the 


positions of the index recorded, a table may be made out as 
follows : 














Average Length of 5 Half Waves . . 
Mean Half Wave Length ... 
Mean Wave Length 

Chladni's Figures 

The nodes of a vibrating stretched string are points of its 
length where, theoretically, there is no motion. On either side 
of the node the string moves simultaneously in opposite directions. 
In the case of a vibrating plate there exist nodal lines, i.e. lines 
in the plate where there is no motion. They divide the plate 
into segments so that the parts on either side of the nodal line 
at any instant are moving in opposite directions. 

The point of support of a plate is necessarily on a nodal line. 
Round and square plates are usually supported centrally in the 
experiment, but interesting results arise when they are supported 

By touching any point on the edge or surface of the plate 
with the finger nail, the plate is prevented from vibrating at the 
point, and if it lies on a nodal line the corresponding mode of 
vibration may be excited. 

In the case of a circular plate clamped at the centre the nodal 
lines are radial, and the fundamental vibration gives two per- 
pendicular diameters. These may be obtained by sprinlding 
white sand over the surface of the plate, touching two points on 



the edge separated by one quarter of the circumference, and 
bowing vertically across the edge at a point one-eighth of the 
circumference from one of the fingers. A square plate clamped at 
the centre gives a large variety of figures. Some of these are 
shown in the figure in order to assist in producing them. 










Fig. 268 

( 14) 

Some of the nodal points on the edge and surface of the plate 
should be touched, and the plate bowed along the edge at one of 
the points midway between the nodes. The fundamental vibra- 
tion gives two perpendicular lines through the centre parallel to 
the edges of the plate. 

The simpler figures are easy to obtain, but skilful manipulation 
of the bow is required to produce the more complicated ones. 


The student should obtain as many as possible, and clamp the 
plate at other points than the centre to investigate other modes 
of vibration. 

The Relation between Pitch and Volume In the ease of a Narrow- 
necked Resonator 

Consider the case of a bottle containing air closed by a piston, 
P, without friction in the neck of the bottle. 



Fig. 269 

Let v denote the volume of the bottle below the piston, m the 
mass of the piston, and A its area of cross-section. Let the piston 
be originally in the position of equilibrium and let the pressure 
outside be p 9 , while that inside is p. Then we have : 

pA = ptA + mg. 

If now the piston be displaced downwards a distance, x, so 
quickly that the change may be regarded as adiabatic, a new 
pressure, p 1 , will be generated, such that : 
p 1 (v — A*)* = pv\ 

We shall suppose that x is small, so that we may write : 

The total force downward is now : 

mg + p<A - p x A == (p - p x ) A 

Thus the equation of motion is : 

d % x pyA'' 


m W = 




This is a simple harmonic motion and the time of vibration 
about the position of equilibrium is : 




(see p. 26) 

In the calculation we have assumed that the pressure is the 
same throughout the gas during the oscillations. 

This is not true since time is required for the transmission of 
the pressure. The other assumption concerning the adiabatic 
character of the compression is very approximately true, es- 
pecially as the neck is narrow and heat will not easily escape 
from the bottle or be transmitted to it. The piston will thus 
behave only roughly according to the formula, and will have an 
approximate period of oscillation given by the above value, 
i.e. it will have a frequency : 

T 27T \ 

£yA s 


Thus we have approximately : 

rih) = constant. 

It is the object of this experiment to verify to what extent the 
gas behaves according to this formula. 

The piston, in practice, is the layer of air in the neck of the 
bottle and by pouring water into it, resonance is produced be- 
tween it and a series of forks of known frequency. A medicine 
bottle will be found convenient and must first be calibrated by 
pouring in water from a measuring flask to various depths and 
measuring the height of the water surface above the bottom of 
the bottle. This should be done for a series of intervals up to 
the base of the neck. 

Make a table thus : 









v n -v x 



f„-t> 2 



v n -v 3 




v n denotes the capacity of the bottle. 

Draw a curve showing the relation between the heights of the 
water and the volumes of air above it, i.e. draw a curve with the 
values h as abscissae and the volumes in column 3 as ordinates. 
By means of this curve we can deduce the volume of air in the 
bottle corresponding to various heights of water. 

Forks of pitch varying from 256 to 512 should be used and a 

curve plotted with volumes as ordinates and values of — as 

abscissae. The resulting curve should be a straight line through 
the origin if the theory is correct, but in practice it will be found 
not to pass through the origin although the relation is very 
nearly linear. The curve will obey closely the law : 

n % (v -f- c) = constant. 

The value of c is a correction to be applied to v. 

This may be regarded as a neck correction, and the ratio of t 
to the volume of the neck should be recorded. 

The agreement of pitch between the fork and bottle should be 
tested by blowing across the neck and noting beats between the 
note obtained and that of the fork. 

The Interference of Sound Waves. Determination of a highly 
pitched Note by means of a Sensitive Flame 

When a continuous note is produced in front of a smooth wall 
the reflected and incident trains of waves produce stationary 
undulation in front of the wall, and consequently there exist 
nodes and antinodes in the air. The experiment consists in 
locating the positions of consecutive nodes from the wall out- 
wards. Thus the wave length is determined, for the distance 
between two consecutive nodes is half a wave length. By 
noting the air temperature and deducing the velocity of sound 
in air appropriate to it (eq. (A), §10 of this chapter), we may 
thus deduce the frequency of the note by the relation : 

V =wx. 

In order to locate the nodes a sensitive flame is used. This 
flame is produced by supplying the gas under pressure to a pin- 
hole burner. The requisite pressure can be obtained by leading 
the gas from the main into a large gas bag and placing a board 
on the top of the inflated bag to carry a weight. 

From the bag a pipe leads the compressed gas to the burner. 
When the bag is full, turn off the main tap and put weights on 
the board so that a tall flame from one to two feet high is pro- 
duced and there is no flaring. It will be found that in the 
sensitive state on making slight noises as, for example, by jingling 

SOUND 425 

keys in the neighbourhood of the jet, the character of the flame 
changes. It flares and shortens, recovering its former state 
when the sound ceases. This is caused by the motion of the air 
as the sound wave passes. 

Start with the flame close to the wall and sound a note from a 
highly pitched whistle, or other suitable source, and move the 
flame slowly outward from the wall. 

It will be found that at certain points the flaring ceases and 
the flame increases in length. At these points the air is still, 
and the points are at nodes of the stationary wave motion. 

If a smooth wall is not conveniently situated, set up a large 
sheet of glass or smooth board at the end of a table, and move the 
flame outward from the surface along the table. The flame will 
probably need adjusting before it will respond readily. To do 
this use the tap leading from the bag to the burner, and also 
vary the weights producing the pressure. It appears to be 
necessary to use a rather long gas tube to convey the gas from 
the bag to the jet. 

In obtaining the most sensitive flame it is to be noted that the 
orientation of the flame is important and the burner should be 
turned about a vertical axis, so that different sides of the flame 
are presented towards the direction of the sound. The flame 
appears to have different degrees of sensitiveness on different 



Measurement of the Pole Strength of a Bar Magnet, using a Grassot 

The search coil of the fluxmeter (see p. 482) is placed on the 
bar magnet as shown in the figure so that it encircles the mid- 
point of the magnet. At this stage the reading of the fluxmeter 
is noted. Then if the bar magnet is uniformly magnetized, 
the coil, when withdrawn, cuts all the lines due to the pole 
past which it moves. 

The fluxmeter indicates the flux change in units which are 
specified ; in the case of the instrument described on p. 482 each 
division corresponds to a change of flux equal to 10000 maxwells. 

If there are x divisions change during the withdrawal of the 
search coil, and there are n turns of wire in the search coil, then 
since from a pole of strength, m, there are 471m lines, we have : 



or m = « 

The experiment should be repeated, using search coils having 

different values for n. 

This gives quite constant values for m as seen in the following 

experimental results. 

Coil A : n = 100. 

Initial reading of Fluxmeter 3 

Final reading of Fluxmeter 46 

Deflection, first experiment 49 

Deflection, second experiment 50 

Mean deflection 49-5 

40-5 x 10000 

m = -^-^ = 394-1. 

47c x 100 J ^ 

Coil B : n = 8. 

Deflection, 4 divisions. 

10000 x 4 

m = -^8 " 3 ^4 

mean value of pole strength, 394*25. 


Magnet was 1*58 cms. by -75 cm., i.e. 1*23 sq. cms. in cross- 
section, i.e. intensity of magnetization, assumed uniform, is : 

394^5 _ 


Distribution of Magnetism along a Bar Magnet 

This may be determined by using the fluxmeter in a manner 
very similar to that of the last experiment. The magnet is 
marked off in centimetres along its length, and the search coil 
is placed at the mid-point, around the magnet. The coil is then 
advanced in centimetre steps and the deflection of the instrument 
noted, i.e. for o to 1 cm., 1 to 2 cms., 2 to 3 cms., etc. The 
deflection in each case is proportional to the magnetization in 
the space moved over by the coil. 

The variation of magnetization along the length is seen by 
plotting deflection against distance from the centre. 

Gauss's Proof of the Law of Force 

The most satisfactory proof that the force between two magnetic 
poles varies inversely as the square of the distance between them 
was first given by Gauss. 

The method consists of a comparison of the magnetic force 
at a point on the axis of a magnet with the force at a point on 
a line drawn at right angles to it, at its mid-point. 

Let us first calculate the value of the force at two such points, 
assuming that the force between poles varies inversely as the 







Fig. 270 

«th power, so that the force between two poles of strengths, 
m and m', r cms. apart is : 


F = 


' End on' position ('A ' position of Gauss) 

Let NS be the magnet (fig. 270) and P a point along the axis 
produced, such that the distance from P to O, the centre of the 


magnet, is r cms. If the length of the magnet is 2.1 cms., we have 
the magnetic force at P, which is the force on unit pole placed 
at P, is : 

m m t, 

.= r A . 

(r - l) n (r + iy 

This is the net repulsion due to N and S on the unit pole at P, 
and is equal to : 

F A =*m 

F =-^ ^ ~ \ ~r) 

If we use a bar magnet such that / is small compared with r, 
we have : 

expanding the two expressions. 

Neglecting the powers of - higher than the first, this expression 

becomes : 

_ 2mnl _ Mw . 

** A r n+l Jn+i' W 

where M is the magnetic moment of the magnet. 

' Broadside on ' Position (' B ' Position of Gauss) 

Again in fig. 270 let Q be a distance r cms. from O, measured 
along OQ, which is drawn at right angles to the magnet at the 

Due to the pole at N, there is a repulsion at Q along NQ equal to 

m , 

{\/(y a + / 2 )}' 

and a similar attraction due to S, along QS. 
The resultant of these forces may be obtained by considering 


the isosceles triangle, QNS, as triangle of forces. The sides 

the resultant 

QS and QN each representing a force ^ 


+ IH*' 



F R = 




f a + / 2 






or when I is small compared with r. 

F --* 

Comparing equations, (1) and (2) we see that for such a short 
magnet the magnetic force at the two points, P and Q, are as n : 1. 

The numerical value oin is obtained by comparing the magnetic 
force at two such positions. 

A magnetometer is employed. This consists of a small magnet 
fastened in a light frame which carries a small mirror, the whole 
being suspended by a thin silk fibre in a cylindrical brass case 
as shown in fig. 271. Fig. 272 shows the usual form of small 
suspended magnet. 

Fig. 271 Fig. 272 

The magnetometer magnet is allowed to come to rest, care 
being taken that all the torsion is removed from the silk suspension. 

Two boxwood scales, C and D, are arranged as shown in 
fig- 273, one in the magnetic meridian, and one at right angles 
to it. 

A lamp, L, and scale, S, are arranged in the usual way, as 
shown, so that an image of the lamp is reflected and focussed 
on the scale S. * The distance of S from the magnetometer mirror 
should be about one metre. 

When the magnetometer needle is arranged to swing freely in the 
centre of the case by means of the levelling screws shown (fig. 271), 
the position of the needle with respect to the ends of the boxwood 
scales can be very readily obtained, if the radius of the cylindrical 
case is measured. 


If now a very small bar magnet is placed at any point, such 
as B, fig. 273, in an E. and W. position, a magnetic field will be 
set up at M, corresponding to the ' broadside on ' position. The 
distance, r, between the magnet centres being obtainable by 


1 1 1 1 1 1 1 1 1 1 u 1 1 ifcgfr 

direct measurement on the boxwood scale. The deflection of 
the magnetometer is found by observation of the movement of 
the reflected beam of light. 

The magnet is then placed, again E. and W., at an equal 
distance from M, on the scale D (A, fig. 273), and the deflection 
again obtained. 

It is desirable to eliminate the zero reading of the magneto- 
meter ; the double deflection, left and right, may easily be 
obtained by reversing the magnet at the two positions A and B. 

The experiment is repeated for various values of r. For 
example, with a magnet 4 cms. long the values of the deflection 
at 50, 60, 70, 80 cms. may be obtained and tabulated as under. 








- = n 












50 cms. 


— l6'7 


8.2 • 






— IO'I 








- 6-2 








- 4-i 








- 2-9 









The deflected position of the magnetometer needle depends 
on the relative strengths of the horizontal component of the 
earth's field, H, and the field due to the deflecting magnet. 
If 6 is the angular deflection produced, we have, 

H sin = F cos 0, 
or F = H tan 0. 

So that for the same value for r 

F\ _ n __ tan t 
F B ~ 1 "" tan 0, ' 
where X is the deflection when the magnet is at A, and 2 when 
the magnet is at B, in the ' broadside on ' position. 

If the distance from the scale S is fixed and the corresponding 
deflections are d 1 and d 2 , 

tan t _ d x 
tan 0, ~ d 2 ' 

d x 
i.e. n = -—- 

This ratio is tabulated in the last column of the table, and a 
mean value obtained. 

The Variation of Residual Magnetism with Temperature 

The object of this experiment is to investigate the behaviour 
of a magnetized carbon steel rod when subjected to temperature 

Fig. 274 

A rod of carbon steel of about 8 cms. in length and -5 cms. 
in diameter is magnetized between the poles of an electromagnet. 
The magnet so formed is set up inside a copper or brass tube, 
which is of slightly larger diameter, as seen in the upper portion 
of fig. 274. 


The magnet is drilled with a sma^lhole which receives a thermo- 
junction, as shown. This thermo-junction is calibrated as 
described on page 545 and serves to register the temperature 
of the magnet. Some simple heating device is arranged to vary 
the temperature of the magnet. In the diagram a second brass 
tube provided on the upper surface with a series of holes is 
shown, fitted on the end of a brass Bunsen burner. If a brass 
Bunsen burner is not available a simple heater may be made 
by taking such a second tube as the one shown, perforated with 
a series of holes, having one end closed and at the other a device 
for admitting an air-gas mixture similar to that of the common 
form of ' gas ring.' 

The magnet is set up at right angles to the meridian, and at 
a distance d cms. from a mirror magnetometer, of the form shown 
in fig. 271 ; the distance, d, should be such that at room tempera- 
tures the magnet produces a full scale deflection. The magnitude 
of the deflection is noted, and the temperature is obtained from 
the thermo-junction balance point on the potentiometer. 

The temperature of the rod is gradually raised by means of 
the heater, and the magnetometer deflection and the thermo- 
junction balance point are noted for every 20 C. rise, until the 
temperature is about i8o°C. At this point the observations 
are made much more frequently as the critical part of the varia- 
tion is being observed. At about 250 C. the readings may be 
again taken at about every io° C. or 20 C. intervals, and observa- 
tions continued until the temperature is at the maximum 
obtainable value for the heater employed. 

The results are tabulated and the value of the temperature 
obtained for each point ; the tangents of the corresponding 
angular deflections of the magnetometer are also tabulated. 

The magnet is then allowed to cool slowly, and the observations 
are continued until finally the specimen is once more at room 

The magnetometer is at a constant distance from the magnet, 
and in a fixed control field, H, hence the moment of the magnet 
is proportional to the tangent of the angle of deflection. 

This quantity is plotted against temperature, and a curve 
somewhat similar to that of fig. 275 is obtained. The firm line 
shows the relation between tan a and t° C. as the temperature 
rises, and the broken line the relationship as the temperature 

For carbon steel in general, it will be found that at about 
200 C. the magnetic moment is reduced to zero. An increase in 
temperature causes a reversal of the polarity as shown in the curve. 
This negative moment is a maximum at about 210 C. and then 
decreases to zero at about 8oo°C. In general such a high 


temperature will not be available with the heater described, but 
the most interesting part of the curve may be obtained. On 
cooling, the negative moment is a maximum at D, at a temperature 
about io° C. or 20 C. above the original position, and on regaining 
room temperature will have a small positive moment, corre- 
sponding to E in fig. 275. 

ioo C 

Fig. 275 

The above experiment which demonstrates an interesting 
feature of the magnetization of such an annealed carbon steel 
rod is due to Prof. S. W. J. Smith and was described by him in 
the " Proceedings of the Physical Society of London," No. xxiv, 
15 Aug., 1 91 2. Reference should be made to that paper, which 
gives an explanation of the observed results in terms of the 
magnetization of the iron and iron carbide molecules which 
compose the bar. 

Briefly, the iron carbide molecules are set in line on magnetiza- 
tion and exert a demagnetizing effect on the iron molecules 
which are reversed in the internal field due to the carbide. At 
about 210 the carbide is totally ineffective, but the reversed 
iron molecules being more retentive have a maximum external 
effect. As the temperature rises, the iron molecules approach 
their neutral temperature, and finally at about 800 entirely lose 
their magnetic effect. (See the original account in the place 
stated above.) 

The Increase in Length 0! a Bar on Magnetization 

To investigate any change in length produced in a rod of iron 
when magnetized, the following apparatus is set up. Fig. 276 
shows a section of a brass case wrapped with two solenoids and 
provided with a water jacket, WW, through which water at a 
steady temperature may be circulated via tubes, T : the two 
coils, C, are wound together so that their magnetic effects 



are identical at the axis of the cylinder. The free ends of each 
coil are connected to separate terminals, P, Q, R and S. The number 
of turns and mean diameter of the two coils are as nearly as 
possible the same, so that if a current were sent through from 







Fig. 276 
P to Q, the magnetic effect is the same as when the same 
current is sent from R to S through the second coil, or if the two 
coils are connected in opposition there would be no magnetic 
field along the axis. 

The method of winding and the resulting magnetic field 
should be tested at the outset by sending a current through the 
coils, and noting the magnetic effect on a compass needle held 
near the end of the solenoid. 

At the top of the brass case is a circular brass end-plate to 
which are soldered rigidly three pins, S, which terminate in 
points on which a sheet of plane glass, G, may rest. 

The iron specimen to be investigated is first thoroughly de- 
magnetized by heating, etc., or by reversals of a diminishing 
current through a solenoid surrounding it (as described on pages 


449-451) . The specimen should be a cylindrical rod with plane ends 
made to fit the cavity provided along the axis of the solenoids. 
The end effects of magnetizing solenoid are overcome by first 
placing inside the apparatus a cylindrical length of brass, B, 
with plane ends, the lower end resting firmly on the thick brass 
base of the instrument. Above the iron rod is placed a second 
brass cylinder, A, also with plane ends ; the upper end projects 
slightly above the level of the upper surface of the end plate. 

A long focus lens is fastened firmly to this brass cylinder by 
means of soft wax or plasticine. The length of the supports, S, 
is such that a small gap is left between the upper face of the 
lens and the lower surface of the glass sheet, G. 

Above the parallel walled sheet of glass, G, is placed a second 
sheet, inclined at 45 to the normal. This reflects light from a 
sodium flame towards the lens and plane surface. 

The reflected beams are viewed by a microscope M, and the 
Newton rings formed are observed in this way. The microscope 
contains a cross-hair which is used as a point of reference. 

When a current is sent through the solenoids arranged in 
series, and to add their magnetic fields, the rod may be gradually 
magnetized by increasing the current strength. If any change 
in length occurs, there will be a movement of the Newton ring 
system, which may be observed in the microscope. 

It will be seen (by reference to page 323), that if the rod 

changes in length by - where X is the wave length of sodium 

light, that one particular part of the field which was formerly 
dark will now be bright or vice versa. The direction of move- 
ment of the fringes will show whether the rod increases or decreases 
in length. 

Now, due to the current circulating in the coils, a certain 
amount of heat will be developed. Thus for a rod of 50 cms. 
length a rise in temperature of about -03 will cause an increase 
in length more than sufficient to move fringes corresponding to a 

change in size of the gap equal to -. It therefore becomes 

necessary to ensure that the rod is maintained constant in 
temperature throughout. Water at constant temperature is 
circulated through the jacket, WW, for some time before com- 
mencing observations and continuously during the experiment. 
Under these circumstances tap water from a supply removed 
from any hot water pipes will be admissible. 

As a further check on the observations, to ensure that the 
results are not spurious due to the heating, each observation is 
preceded by one in which the current to be used is sent through 


the twp coils arranged in opposition, i.e. the heating which may 
affect the dimensions of the bar will be present, and any movement 
of the fringes is due to that cause, since there is no net magnetic 

The coils are rapidly changed over, and the same current 
causes a magnetization of the bar. The true effect on the 
length of the bar due to magnetization may therefore be obtained 

To facilitate the rapid change in the direction of the current, 
a Pohl commutator may be employed : the ends of one coil, 
say, R and S, are connected to one pair of the terminals of the 
commutator ; the other coil is connected in series with a battery 
and variable resistance to the middle terminals. Thus by throwing 
over the movable arm, the coils may be changed from a condition 
of producing fields in opposition to one in which their magnetic 
fields are in the same direction. 

Carry out observations using currents up to a maximum value 
which is determined by the current-carrying capacity of the 
coils, CC. 
• Plot the results showing change in length in terms of the 
wave length of sodium light, against the field strength measured 
in Gauss. 



Determination o! the Horizontal Components of the Earth's Magnetic 
Field (H) 

The method due to Gauss, described below, is usually employed 
to determine the horizontal component of the earth's magnetic 
field ; it may also be used to measure any magnetic field which 
is uniform over a sufficiently large volume. 

The method involves two experiments. In the first a magnet 
of known moment of inertia is suspended freely in the earth's 
field, at the place where H is to be found, and from observation 
of the time of swing, the product MH where M is the moment 
of the magnet, may be calculated. In the second experiment 
the field due to the same magnet is compared with the earth's 
field by means of a magnetometer. 

Determination of MH 

Let the magnet be suspended in a light stirrup and perform 
oscillations whose periodic time is T seconds, it being supposed 
that the suspension has no initial twist. Then if I is the moment 
of inertia of the magnet about the axis of suspension, and i the 
moment of inertia of the suspending frame about the same axis, 
and if t be the restoring couple per unit angular displacement 
due to torsion of the suspending fibre, we have : 

= 2*J 

'(MH + t)' {1) 

for (I + i) is the moment of inertia of the system, and (MH + t) 
is the total restoring couple per unit angular displacement, when 
small displacements are considered, whence : 

MH = A, (say) (ia) 


To find jr 

A magnetometer is set up in the place at which MH was 
found, and the magnet placed with its centre d cms. from the 
centre of the magnetometer needle, east or west of it and lying 
east and west. If be the deflection of the magnetometer, 
and F the value of the field due to the magnet, we have : 



F = H tan 0, 

and F = ^ 2 _j 2)2 > 

where il is the distance between the poles of the magnet. 

Alternatively, if the magnet were placed so as to be in the 
€ broadside on ' position, i.e. E. and W. with its centre in the 
meridian through the centre of the magnetometer needle, we have 
F lf the field produced at the magnetometer, in this case is 

~ (d t * + /*)*' 
where d t is the distance between the centre of the magnet and 
the magnetometer in this case, and I is the distance between the 
poles of the magnet. 

Whence for the ' end on ' position 

Htan6 = (i2 _ /2 p (2) 

or for the ' broadside on ' position 

Htan^ = p^47iji (3) 


in either case the values of g may be found. 

Let this value be B. 

Whence from the two experiments : 



In performing these experiments special forms of magnetometer, 
etc., are advisable. All iron is removed from the neighbourhood 
at the outset, and care taken to maintain the magnetic conditions 
the same throughout the two experiments. 

The magnet used may very well be a true cylindrical bar 
magnet of not more than a few centimetres in length. 

Let the length as determined to & of a mm. be 2l x cms., and 
the radius, measured with a micrometer screw, be r cms, the 
mass of the magnet, weighed to at least one in a thousand, being 
m grammes. I, its moment of inertia about an axis through 
the centre of gravity and normal to its length, is 


To find T, use is made of a light stirrup suspended by a fibre 
from a torsion head. A brass cylinder of the same mass as the 
magnet is first placed in the stirrup, and all the twist is then 
removed from the fibre. The torsion head is then turned so 



that the brass rod is in the magnetic meridian. It is then 
replaced by the magnet. 

The frame is provided with a small mirror, otherwise one end 
of the magnet is polished to act as a mirror, and a lamp and 


Fig. 277 

scale is arranged so that a beam of light reflected from the 
mirror on to the scale enables the movement of the suspended 
system to be observed. 

Set the magnet swinging through a very small angle, and 
obtain the time taken for 100 complete swings. Repeat this 
and take a mean value, from which T may be obtained.* 

To find i in equation (1) weigh the stirrup and estimate the 
moment of inertia from the dimensions. 

The restoring couple due to the torsion, t, may be obtained 
by rotating the torsion head through, say, a complete turn. 
Note the resulting angular deflection of the magnet, y> radians, 

whence (2* — y>) t = MHy, 


or T = 7 r- 

(27C — y>) 

It will be found that, if the fibre is thin and the stirrup a very 
light one, t and i are negligible compared with MH and I. 
Including these corrections we have : 

T = 2* 


^H(x +_£_)' 

* Alternatively, the method of timing set out on page 118 may be used. 




4* 2 (I + i) 

T 2 ^r~ 

27t — \p 

The magnetometer experiment is performed, using a special 
form of magnetometer (Kew type magnetometer). This consists 
oi several small magnets rigidly fastened in a metal frame, 
which also carnes a small mirror. The whole is encased in a 
brass case, provided with suitably arranged glass windows so 
that a beam of light may be reflected from the mirror on to a 
scale S cms. away from it. 

The magnetometer frame carries a long bar which has four 
pegs, Pj, P 2 , P 3 , p 4) so arranged that when a small carriage is 
placed on any of them, the cylindrical magnet, which the carriage 
supports, lies with its axis in the same plane as the magneto- 
meter needle. Figs. 278 and 279 show diagrammatically the 
arrangement of the magnetometer. 





"i 1 1 1 1 ri 1 1 1 1 

Fig. 278 




Fig. 279 

The magnetometer and bar are first levelled (using a spirit-level) 
by means of the levelling screws on the base of the instrument. 

Pj and P 4 , P 2 and P 3 are fixed, in the construction of the 
instrument, equidistant from the needle when the system is 
level. Measure P X P 4 = D x cms., say ; measure P 2 P 3 = D 2 cms. 

The scale is then adjusted so that the reflected spot of light 
is at its centre when the magnetometer is under the influence 
of the earth's field alone. The arm which carries the pegs is 
set first of all in the direction of the magnetic meridian, and the 
small carriage in which the cylindrical magnet is carried is 
placed on V x and turned round so that the magnet is at right 
angles to the meridian. The true position for the magnet at 
right angles to the meridian is obtained by noting the deflection 
of the magnetometer. This will be a maximum when the normal 
position is acquired. 


In this manner the deflection in cms. on the scale may be 
measured. The magnet is then rotated through 180 until a 
maximum deflection in the opposite direction is obtained in the 
magnetometer. Let these deflections be <5 X and <V cms., and the 
corresponding deflections when the magnet is supported on the 
other pegs be 6 2 , 6 2 ', etc. 

Obtain the mean value of 6 t d t ', <5 4 6 J — 6 say, that is the 
mean deflection for distance £D X — d lt ; also take the mean of 
#2 <V> ^3 <V — sl > sa Y> corresponding to a distance £D 2 = d 2 . 

If q> and <p' are the corresponding angular deflections of the 
magnetometer, we have : 

St St 

tan 29 = ^- ; tan 29' = ~-» 

Hence we may calculate q> and q>' from the observed deflections. 

The cylindrical magnet was placed as described above so that 
the field at the magnetometer was that due to the ' broadside on ' 
position ; thus from equation (3) above 

H tan * = w+w 

i.e. g = (df + l*f • tan g> = (d 2 * + If tan 9'. 



(i» + jy = rf»(i+!j!-. . .), 

neglecting the fourth and higher powers of -5. 

i.e. H = d, 3 

I+ !(j,)^ tan "' 

M ■•'^.fCr i )fw. 



Hence / may be eliminated, giving 

( * - * A =d 2_ d , 

V^tany d 2 tan <p'J x 2 ' 

Hence =5. is determined, and the value of H may be calculated 

using this and the 1 previously obtained value of MH. 

The bar carrying the four pegs may be alternatively arranged 
at right angles to the meridian, and the carriage containing the 
magnet again placed on the pegs with the magnet at right angles 
to the meridian, producing a field of force at the magnetometer 
corresponding to the ' end on ' position. The process described 
above is repeated, and the average deflections obtained enable 


the calculation of the mean deflections, x and 2 , for the distances, 

d x and d v 

Hence again : 

TTi „ 2M I 

H tan X = ^r-T r x 
D x 3 

2£ 2 ) 

d^ ' ' '] 

TT A „ 2M 
Htan 2 = ==r-. X 

D 2 S ( T _ i 2 ) ' 

to the same order as before. 

Once more eliminating I, half the distance between the poles 
of the magnet, we have 

2 .M(_I j * ^ =^-^, 

H V*! tan ©j d % tan 2 / 


whence ^ may be again calculated. This second method gives 
larger deflections and is therefore preferable to the first. 

Fig. 280 

The Dip Circle— Measurement of the Angle of Dip 

The dip circle consists of a long magnet supported on a hori- 
zontal axis which passes (approximately) through its centre of 
gravity and is at right angles to its length. The magnet is sup- 


ported in wheel bearings or on agate edges, so that it may turn 
freely and with a minimum of friction. On the same base as 
the support for the knife-edges is a vertical graduated circle 
which enables the position of the ends of the needle to be 
obtained. The whole structure is supported on a circular table, 
which is capable of rotation around a vertical axis. This rotation 
may be measured on a horizontal circular scale about which the 
table may rotate about its axis, and, by means of a suitable 
vernier, may be measured to at least one minute (see fig. 280). 

The position of the ends of the magnet may be read on the 
vertical scale by means of microscopes carrying verniers, which 
move round the scale. If the plane of the vertical scale be turned 
in the direction of the meridian, so that the horizontal axis of sup- 
port of the magnet is at right angles to it, the magnet sets in the 
direction of the earth's lines of magnetic force, and the angle 
included between the horizontal and the position of the needle 
is the angle of dip. ' 

However, there are many sources of error . The axis of rotation 
/ . of the magnet may not quite coincide with the centre of the 
scale. One end, therefore, of the magnet would read too small, 
and the other too large a deflection. 
2 The axis of rotation may not be truly through the centre of 
" gravity, in which case a couple is exerted, tending to turn the 
magnet so that the centre of gravity comes vertically under the 
axis of support. 
^ The axis of the magnet does not usually coincide with the 
geometric axis of the needle, so that the centres of the ends of 
the magnet do not give a correct reading for the magnet when 
set along the earth's line. The possible errors are fully set out 
in any textbook of magnetism. 

A consideration of the observations taken and described below 
will show that these errors may be eliminated. 

To determine the angle of dip, the dip circle is set up and 
by means of the levelling screws on the base of the instrument, 
using a spirit-level, the 'horizontal ' scale is made truly horizontal. 

The plane of the meridian must then be found, so that the 
magnet may set freely. To do this, the instrument is first 
turned so that the upper, S, pole is at the 90 mark, and the 
reading of the vernier on the horizontal scale is noted. 
• It will in general be found that the lower, north, pole of the 
magnet is not quite at the lower 90 scale reading. A slight 
turn of the screw which rotates the instrument about the vertical 
axis is then made until the north pole is at the 90 reading, and 
the horizontal vernier scale is once more read. The circle is 
next turned through 180 , and the two more observations are 
made. The needle is now reversed in its bearings and the four 


observations repeated. The mean value of the horizontal scale 
readings for these eight positions is taken. This is the mean 
position for the plane of rotation at right angles to the meridian. 

The case is then moved through an angle of 90 from this 
mean position so that the plane of the needle now coincides with 
the magnetic meridian. 

The following 16 readings are then obtained : 

(1) Read the position of each end of the magnet. 

(2) Rotate the instrument through 180 , and once more read 
both ends of the magnet. 

(3) Reverse the needle in its bearings and read both ends on 
the graduated circle. 

(4) Turn the instrument through 180 and again read both ends. 

(5) Remagnetize the needle and repeat the above process. 
The needle is remagnetized so that its poles are intercharged. 
The north pole for the observations, 1 to 4, above now becomes 

a south pole. Under these new conditions eight more values 
are obtained. The mean value of the 16 readings gives the 
true value of the inclination. The observations may be con- 
veniently tabulated as under. 







magnet remagnetized 
(needle readings) 











Needle reversed in bearings. 





q+ P + r + 8 

== A 

Angle of Dip = 


A +B 

a' + P' + Y' + <r 

= B 

Note. — It may be necessary to obtain corresponding 16 values for the 
determination of the zero instead of the 8 described on pages 443-4' 


Measurement o! Permeability by the Magnetometer Method 

Consider a long thin specimen of iron wire placed in a magnetic 
fiield of uniform strength H. Due to the fields the soft iron 
specimen becomes magnetized, having poles of strength, ± m. 
Let , 

2/ be the distance between the magnetic poles, 
s be the cross-section of the wire. 

The number of lines coming out from the north pole end is 
47cm, and an equal number enter the south. Due to the poles 
developed, there are a number of lines of magnetic force passing 
from the south to the north in the specimen, equal to ^nm. 

This is equivalent to - — per sq. cm. of cross-section of the 

iron, assuming that the lines enter and leave at the ends only. 
In addition to the above there are H lines per sq. cm., due to 

the field, making the total number per sq. cm. of H + - — • 

This is the induction, B, in the specimen. Thus N 

The magnetic moment of the wire is 2,1m, its volume is s • 2/. 
Hence the intensity of magnetization, which is the magnetic 
moment per unit volume, is 

2.1m _ m _ T 
2.1s ~ s ~ 

Hence B = H + 4* 1 ( x ) 

The permeability (v) for any value^of H is defined by the ratio — • 
We therefore have : 

[L = 1 + 4« jj 

The ratio -=y is usually defined as the susceptibility or coefficient 


of induced magnetization, so we have, putting k for this quantity, 

jx = 1 -f- 4nk. 


We have imagined the length of the magnet to be large. Other- 
wise there is an effect due to the poles at the end of the specimen 
to be considered, in expressing the value of H. For, due to the 
north and south poles developed in the specimen, there will be 
a magnetic field in the opposite direction to H, reducing the 
value to H 1 , say. This demagnetizing field can be neglected, 
however, if 2.1 is made large compared with s. A suitable length 
is obtained when 2X is greater than 200 times the diameter of the 

If the specimen is arranged horizontally or vertically near a 
small magnet, suspended by an unspun silk fibre, the magnet 
will be deflected by the induced magnetism in the specimen. 
From a knowledge of the deflection and of H, the value of the 
constants, B, \k, I, k, may be calculated. 

To find the above constants for a given wire specimen, the 
type of reflecting magnetometer described on page 429 is used. 

The inducing magnetic field is obtained from a solenoid, 
through which a current of known magnitude may be passed, 
and the deflection of the magnetometer may be observed. 

Iron Wire Specimen arranged Vertically 

The arrangement of the apparatus is seen in fig. 281. L is 
a lamp and scale, M the magnetometer which is clamped to the 
wooden base shown, at a convenient distance from the specimen 
which is placed in a vertical solenoid, S. The wooden base is 
arranged in an east and west direction so that the magnetic 
effect of the specimen may be measured. 

Fig. 281 

Before introducing the specimen, the solenoid is connected 
in series with a small coil, C, and several accumulators, a variable 
resistance, an ammeter reading to 3 amperes, and a reversing 

The coil, C, is so arranged that the magnetic effect produced 
by it on the magnetometer is the opposite sign to that produced 
by the solenoid, S. 


The position of C is adjusted so that whatever is the value 
of the current in the circuit there is no movement of the needle. 
In that way the whole movement of the magnetometer will be due 
to the magnetism induced in the iron wire. 

The current is then switched off and, having noted the position 
of the reflected spot of light on the scale, the unmagnetized 
specimen is introduced. If there is any movement of the spot 
of light it may mean that there is a certain amount of magnetiza- 
tion in the specimen. It should be demagnetized by first of all 
heating to red heat and then hammering vigorously. Having 
once more replaced it in position in the solenoid, the current 
should be adjusted to about one ampere. The direction of the 
current should then be constantly reversed, whilst at the same 
time its magnitude should be reduced to zero. In this way 
any small residual magnetization may be neutralized. 

A preliminary experiment on a second sample of the same iron 
wire enables a value of the maximum current to be determined. 
The distance MS is adjusted so that a full scale deflection is 
obtained when the sample is saturated. The original specimen 
is then introduced so that its lower end is level with the axis 
of the magnetometer magnet. 

Starting with the soft iron unmagnetized specimen, the zero 
position of the magnetometer on the scale is observed. The 
value of the current is then raised to, say, '2 ampere and the 
deflection noted. Then without breaking the circuit the current 
strength is increased by increments of m 2 to a final value of about 
3 amperes, a limit which depends upon the thickness of the wire 
in the solenoid winding, and also upon the result of the preliminary 
experiment, whichever limit is the smaller fixes the maximum 

Having increased the current to a suitable maximum value, 
it is then decreased by similar steps, again taking care that the 
circuit is not broken in the process, until zero current circulates. 
The current is then reversed and increased to the same maximum 
in the other direction. The current is again brought back in steps 
to zero, and having once more reversed the current it is raised 
to the maximum in the original direction. 

Tabulating these results a connexion is found between the 
value of the current in amperes and the magnitude of the deflec- 
tions or the value of the scale reading, which, when plotted, will 
show the general form of the curve below. 

The results are then converted to the corresponding C.G.S. 
values for B and H. 

In fig. 283 let n s represent the magnetized specimen which 
will produce at P the position of the magnetometer, a field 


strength, F, at right angles to the Earth's horizontal component, 
and which causes the deflection, 0, say. 

We have F=H'tan0, 

where H y is the magnitude of the Earth's horizontal component. 

Fig. 282 

Let m be the pole strength induced in the specimen, 
I be the length in cms., 

r the distance from the lower end of specimen to the 
magnetometer in cms. 







u» r - 

Fig. 283 



F = — — ^r-. cos SPn 
r* SP 2 



r 2 r 2 + /* (r 2 +l 2 )* 

/i r \ 

V 2 (r 2 +/ 2 )v 


Whence : 


m = 

/L I \ ' 

\y2 ( y 2 + l^ij 

and I = -, B = H + 47rl. 

Further, using a solenoid of n turns per cm. having a 

current of c amperes, the uniform field H = - — is effective on 


the specimen. 

Hence, for each reading in the table of results, I, B and H, 
also (x and k, may be calculated. If the curve is drawn for I and 
H, the area enclosed is equal to the work done on the specimen 
in the cycle (i.e., the heat developed). 

The values of H and B are in C.G.S. units, i.e. Gausses (see 
table on page 633). 

(2) The B-H Curve for a Sample of Iron (using a Ballistic Galvanometer 

This method is specially applicable to the determination of 
the B-H curve for a specimen in the form of an anchor ring, or 
a very short hollow cylinder. For such a specimen, magnetized 
by a magnetic force of, say, H gausses, no free poles will be 
developed, and therefore no demagnetizing field will be set up. 
The value of the induction, B, will therefore correspond to H, 
and not some smaller field, H', as in the previous case. 

The experimental arrangements of this method, as shown in 
fig. 284, are such that a variable field, H, may be set up, by 
passing a current through a primary winding, P, which is closely 
wound on the anchor ring, and some means of measuring B in 
the specimen. 

The method employed to measure B is to wrap a few turns 
of wire, S, round the anchor ring and primary winding and 
measure the quantity of electricity which passes through a 
ballistic galvanometer, BG, due to the change in the induction 
in the specimen for a known change in H. Since the ballistic 
galvanometer must be standardized, the secondary circuit is 
completed through a second small coil of a mutual inductance, 
M, used in the standardizing experiment. Thus the ballistic 
galvanometer is in a fixed resistance circuit in all measurements. 

The current from an accumulator may be regulated by resist- 
ances R x and R 2 and measured by an ammeter, A. By means 
of the Pohl commutator the current may be sent directly or 
reversed through the primary windings, P, when K 2 is to the 
left, or through the standardizing mutual inductance, M, when 
the key K 2 is closed on the right. 


By closing the key K lt the resistance R 2 may be cut out. 

The value of H may be calculated, as explained later, from 
the value of the current strength as obtained from the ammeter ; 
and B may be calculated from the observed throw of the ballistic 

E -p; 

Fig. 284 

galvanometer, BG. Since the galvanometer circuit is composed 
of a comparatively low resistance, a moving coil instrument may 
be too heavily damped to be serviceable (see page 481). A 
moving needle instrument is used if this is the case. 

As a preliminary experiment, the key K 2 is closed to the left, 
and R x and R 2 decreased, until, on closing the Pohl commutator, 
the galvanometer gives a full scale deflection from the zero, 
which should be the central graduation of the scale. The 
current required to do this is noted and is used as a maximum 
value in the main experiment. Usually from 2 to 3 amperes 
will suffice, the sensitivity of the galvanometer being adjusted 
to measure the throw produced. 

In general, the past history of the iron anchor ring will be 
such that residual magnetism in the specimen is almost certain. 
This must now be reduced to zero. 

To do this the galvanometer circuit is broken and the resistances 
R x and R 2 reduced to a minimum. The current passing in P 
is then reversed many times and R x and R 2 gradually increased 
until finally the current which is reversed is very small. It 
should be noted that R x and R 2 are such that changes in resistance 
may be brought about without breaking the circuit. 



Another method of demagnetizing the specimen is to pass an 
alternating current through P and a liquid resistance in series. 
The alternating current is gradually reduced to zero by with- 
drawing the electrodes of the resistance. 

The galvanometer is once more put in circuit with S, etc., 
by closing the key in the circuit (not shown in the diagram) : 
K x is closed, and R x given the value corresponding to the maxi- 
mum current as determined by the preliminary experiment. 
The commutator is then closed to the right, and the throw of the 
ballistic galvanometer, <5, noted. The reading of the ammeter 
is also noted. The throw will correspond to an induction, Bj, 
and the ammeter reading to a magnetizing force, H x , represented 
by some point such as S in fig. 285. We now use this as a point 
of reference. The galvanometer circuit is again broken, and the 
commutator is reversed rapidly some 20 to 25 times, and finally 
left on the right, i.e. the iron is taken several times round the 
cycle represented in fig. 285 and is said to be in the 'cyclic state.' 
BG is again put in circuit. When all is steady, R 2 is given a 
small value and K x is opened. The magnetizing field is thereby 
decreased and the throw of the galvanometer, d 2 , noted. This 
throw corresponds to a decrease in induction, Bj — B 2 . The 
value of the current corresponding to H a is noted on the ammeter. 

Fig. 285 

Kj is now closed, the galvanometer key opened, the Pohl 
commutator reversed 20 to 25 times, and finally left to the right. 
The galvanometer is put in circuit ; R 2 is given a larger value ; 
K x is then opened and the throw due to the change in the induc- 
tion, say, Bi — B 3 , is noted ; the ammeter is again read. 

This process is repeated with the commutator to the right, 
until R 2 is infinite and consequently the current and H zero, 
i.e. the relation of B to H represented by the path, SA, of fig. 285, 
is investigated. 

After each measurement, the iron, by the reversal of the 
maximum current, is returned to the state represented by S, 
which therefore becomes the reference point. 


The key, K lf being closed, etc., the commutator is reversed 
some twenty times once more and finally left to the right ; R 2 is 
given a large value and the galvanometer is again put in circuit. 

The commutator is then thrown over to the left, and at the 
same time K x is opened, i.e. the current is reversed and at the 
same time made of small value. This gives a point on the part 
AB of the curve. The same starting point is taken (S), the 
drop in induction measured, and the change in H from a maximum 
position to a small negative value obtained. 

This process is repeated in many steps until finally R 2 = o, 
i.e., until the change in field, H, and induction, B, corresponds 
to a reversal of the maximum positive to maximum negative 

At this stage we may assume that the curve is symmetrical, 
and obtain the full hysteresis loop having obtained SABS 1 
experimentally, or we may proceed to find SfA^S by experi- 
ment, using S 1 now as our reference point. The process is as 
before except that the commutator is now left on the left-hand 
side. K x is opened in turn to R 2 having values from o to infinity. 



















from S 


to -24 = 2-o cms. 



•24 to -22 = 1-5 cms. 




•24 to «20 = i*8 cms. 




•24 to- 18 = 2-4 cms. 



. . . 




— 02 

— 20 

— 22 


•24 to — 24 = 

20000, say 

— 10000 


At this stage (point A 1 ) R 2 is given values from infinity to o. 
and Kj opened as the commutator is thrown over to the right- 
hand side. 

It should be found that the value of the throw for the reversal, 
when the full current is passing through the primary is identical 
for the change from S to S 1 and S 1 to S, and that either value 
is approximately twice the first recorded throw, <5 X . 

The results may be tabulated as shown. The original direction 
of the current in the coil, P, is taken as positive and the current 
value may be recorded in column (1) and the corresponding throw 
in column (3). 

To Calculate H 

Imagine a unit magnetic pole to be taken round the axial 
circle of the solenoid. If H is the value of the magnetic field 
strength due to a current, C, in the solenoid, the work done on the 
pole for such a complete circular path is 2i*r • H, where r is the 
mean radius of the anchor ring. 

If there are N turns in the winding, the unit pole is linked 
with each winding in the complete path, therefore doing work 
47rC for each or a total of 4nNC, 
e.g. 471NC = 2TtrK 

H = ?^,* <D 


where C is the current strength in E.M. units (not amperes). 

The value of the current in E.M. units (not amperes) shown in 
column (1) may be converted to gausses (lines per sq. cm.) by 

multiplying by the factor — as seen in equation (1). 

To calculate the induction, B, corresponding to the observed 
value of 0, an auxiliary experiment is necessary. The two-way 
switch, K 2 , is closed on the right, so that a current maybe sent 
through the long straight solenoid, M (about 40 cms. long and 
4 cms. in diameter), which has m 1 turns per cm. Inside M is 
the second small solenoid which has been in series with S and 
the galvanometer throughout the preceding experiment. 

For a current of C E.M. units flowing through M, since the 
solenoid is long, the field strength at the centre is ^imfi gausses. 

*The result of (i) may be obtained from the general formula giving H inside 

a solenoid of n x turns per unit length, i.e. 

H = 4irtt]C. 

In our case n x — -—-,> 

H r 




Let the radius of the inner coil be r 2 cms., and N a the total 
number of turns in this coil, then the total flux in the inner 
coil is : 

4izm 1 C(izr 2 e N 2 ) maxwells. 

If now the current be reversed in M the change in induction 
in the central coil is : 

8n*m 1 Cr t , N 2 lines. 

Let R be the total resistance in the galvanometer circuit. 

Since the electromotive force is numerically ~rr# 


r> <*N f ,, fdN 

Q = **n£*S mC (2) 

For an instrument of the moving-needle type : 

e = *G sm 2 ~( I+ S> 

Q = K 1 sini( I+ ^). 

T H 
where K t = — . ^-and is constant. 

7C (jr 

Now for small deflections, if d is the scale deflection in cms. 


sin oc d oc sin - and x is constant in the present case, even using 

a moving coil instrument, for the circuit is constant, 

i.e. Q = K 2 i; 

where K 2 is a constant (i.e. Qccd for either type of galvanometer) 
substituting in (2) above 

^ = iL^a, 

or Kd = Stt^jN^C, 

where K = K 2 R == a constant. 

The last equation, which connects the deflection, d, with the 
total change in flux, enables the calculation of the value of B 
corresponding to each field H. 

The change in the number of lines, threading the entire circuit, 
corresponding to 1 scale division deflection on the galvanometer 
scale is : 

K ^ d 


The deflections 8 V <5 2 , etc., in column (3) therefore correspond, 
to a change in flux equal to Kd lt K<5 2 , etc. This is a change in 
the total flux in the space of the secondary winding S. If r x is 
the radius of cross-section of the iron anchor ring, approxi- 
mately that of the secondary winding, S, and m is the number of 
turns in the secondary, and (Bj — B 2 ) in the change in induction 
to be entered in column (4), we have, since B is the number of lines 
per square cm. of cross-section : 

(B x - B 2 )m • Ttf-j 2 = K<5, 

Bl-B « = ^'* 

( a m r x 2 ) 

The term in the brackets is evaluated and used as a reduction 
factor, converting 6 to induction for column (4). 

To obtain column (5), assume a symmetrical form for the 
curve, i.e. assume that the induction for the maximum positive 
and negative current values are equally placed from the zero 
induction line, i.e. equal to half the induction for a full current 
reversal, which is the mean of the extreme values in (4). The 
method of calculation may be understood from the numbers shown 
in the suggested table. 


The B-H curve could also be obtained for an anchor ring as 
above, using a fluxmeter in place of the ballistic galvanometer. 

The calculation of H is as before. When the induction is 
large the number of turns in the secondary coil can be diminished, 
so that the fluxmeter reading is not excessive. The calculation 
of B from the observed deflection, 0, for any current change in 
the primary, is more direct than for the ballistic galvanometer. 
Unit scale deflection on the fluxmeter corresponds to a flux of 
10000 maxwells. So that if the secondary coil used is of m 
turns, and the radius of each turn of the secondary is r lt the 
induction change (Bj — B 2 ) corresponding to the deflection 0, 
is given by : 

WX(Bi — B 2 ) Xtt/! 2 = 100000, 

„ -. 100000 
i>i — Jt>2 = r« 

The experiment is carried out in a manner similar to that 
described above for the galvanometer as flux measurer. 



The most convenient form of current measurer is the ampere 
meter or ammeter, which indicates directly, on a graduated scale, 
the strength of the current passing through the circuit in which 
it is placed. 

As a result of the method of construction the range of useful- 
ness of the instrument is limited, in the usual form, to measure- 
ment of current not less than a milliampere. Recently, however, 
by improving the suspension, etc., the two-pivot instrument 
has been manufactured capable of measuring one microampere, 
for example, a Weston ammeter is made with an open scale 
division equal to one microampere. 

Similar conditions hold for the voltmeter. 

Fig. 286 

The Voltmeter 

This instrument consists essentially of a coil of thin copper 
wire, C, which is supported on an axis pivoted on jewelled pivots, 
P, P 1 , and is free to move in the cylindrical gap between the 
soft iron pole piece, DD\ of an aged steel permanent magnet, NS. 



The field is strengthened and made approximately radial by 
the insertion of a soft iron cylinder; I, in the space inside the coil. 
This cylinder, as Seen in the lower fig. 286, is fixed to a brass 
bar, B, which forms a bridge between the pole pieces. The 
air gap is thus reduced and the coil moves freely in this 

To establish a restoring couple, the two hair-springs, H, H 1 , 
are attached to the axis, and to points on the framework which 
are insulated from each other. These springs also serve as 
leads for the current, to and from the coil. 

Attached near the upper spring is a light counterpoised pointer 
which moves over a scale G. The centre of gravity of the whole 
system is arranged to coincide with the axis of suspension. 

When in use the instrument is placed in parallel with the 
points whose potential difference is to be measured. The 
internal resistance of the instrument must, therefore, be large 
in order to avoid any appreciable rearrangement of current and 
potential drop in the circuit. The current passing through the 
voltmeter is therefore very small for such a high internal 
resistance, and therefore the heating of the coil is not very great. 

The internal resistance is not made up entirely of that of the 
copper coil. In most forms the greater part of the internal 
resistance consists of a resistance in series with it (fig. 287). The 
chief reason for this is to avoid any error due to heating in the 
moving coil. 

Such heating may be due to (a) atmospheric rise in temperature, 
( p) the Joule (C 2 R) effect. An increase in resistance of the moving 
part would occur in either case unless the temperature coefficient 
of the wire were small. Manganin has a small temperature 
coefficient, but a higher specific resistance, i.e. for the same coil 
resistance a less radiating surface is available for manganin 
than copper. 

The effect of (a) in raising the temperature of the coil may 
be best eliminated by making the moving coil resistance fairly 
low, and hence the percentage change of the whole is 

Hence the usual compromise is a copper moving coil of com- 
paratively small resistance and a series manganin coil of high 
resistance. This series resistance is constructed of thicker wire 
than would be possible for the moving coil. 

Having in this manner secured the best approximation to 
constancy for the internal resistance, it will be seen that the 
small current, C, passing through the instrument is proportional 
to the potential difference between the terminals. Now the 
deflection produced is proportional to the current for a large 
range of deflection, when the field is radial, and hence the deflec- 


tion produced is proportional to the potential difference between 
the terminals. 

The instrument is made ' dead beat ' by winding the moving 
coil on a copper frame. 

Seia Rfesisra^ce 

To >->pPer >b \ouo«r v 



-3 Volts 

Fig. 287 


-»50 Volte 

The same instrument may be used to measure different ranges 
of potential. This will be seen from fig. 287. Thus if a potential 
of 3 volts, when applied to PL, produces a full scale deflection 
in the instrument, whose internal resistance (AB + coil) is 
345*5 ohms, it will be seen that the current in the coil is -00865 

Now, if a higher potential, say, 150 volts, is to be the new 
value corresponding to a full scale deflection, this may be con- 
nected to H and P so that a bigger series resistance, BC, is 
included, such that the total internal resistance, R, is given by 

•00865 = ^, 

or R is 17270 ohms. 

The same current will flow through the coil and hence the 
deflection will be again a full scale deflection. 

The graduations on the scale will therefore subdivide the 
o to 150 into equal increments, and each division corresponds to 
fifty times the value which corresponds to the lower voltage 
applied to PL. 

For the lower range voltmeter, measuring potential of the 
order of a millivolt, the value of the internal resistance is 
smaller, for the deflection is proportional to BC, where B is the 
magnetic flux in the gap and C the current. B is constant and 
so C to produce the deflection, when a low potential is applied, 
is obtained by decreasing R. 


In such a moving-coil instrument the direction of deflection 
depends upon the directions of the current. The higher potential 
should be connected to P which is marked +. An accidental 
reversal of this order is apt to strain the needle. 


To measure the current in a circuit, the measuring instrument 
used should be of low resistance, unless some account be taken 
(as in galvanometers) of the resistance introduced in this way. 
For example, if a small resistance of known magnitude, r, be 
included in the circuit, the current strength, c, may be found 
if the value of the potential drop, V, along r, be determined by 


means of a milli voltmeter, for c = — . This arrangement of a 


voltmeter shunted with a low resistance is utilized in the am- 
meter. The fixed-range ammeter usually contains the shunt 
inside the case. The value of the shunt resistance is small, 
and therefore the resistance of the whole instrument is of the 
same order. The moving coil of the ammeter is often, also, 
provided with a series resistance as in the voltmeter to minimize 
temperature variations as described above. 

Many of the better forms of ammeter are not provided with 
fixed shunts but require an external one. The value of the 
shunt resistance determines the range of the instrument. 

For bigger ranges the smaller is the resistance of the shunt. 
Suppose, as before, the maximum scale reading is obtained for 
a current, c, through the coil ; this is proportional to the potential 
difference between the ends of the shunt. It will be obvious, 
that if the external current, C, be doubled the drop of 
potential along the shunt will be doubled, so that if a shunt of 
half the original resistance replace the first, the potential drop 
will be equal to that which is required to send a current, c, through 
the coil and series resistance of the instrument, and so produce 
a full scale deflection. 

The shunts are made of manganin, which has a low temperature 
coefficient. The dimensions required in using a definite manganin 
strip may be calculated. If it is found that to produce no 
appreciable heating the shunt width has to be excessive, it is 
usual to construct the shunt of several strips in parallel. 

The instrument should never be used without the appropriate 
shunt for the current to be measured. 

The ammeters and voltmeters described above have the 
advantage of being direct reading on a calibrated scale ; they 
are robust and do not require any adjustment. But as indicated 
at the outset, the general type of instrument is not sufficiently 
sensitive to measure currents of less than a milliampere or 


potential less than one millivolt ; but by more delicate con- 
struction and general improvement they can be made to measure 
to one micro-ampere and micro-volt. In such a case the extra 
sensitivity entails^ very precise work, and makes the cost of the 
instrument somewhat higher than for the ordinary range (i.e. 
one millivolt or ampere). 

Some forms of instrument are available which combine the 
voltmeter and ammeter. The necessary shunts and series 
resistance are contained inside the case, and by connecting to the 
proper terminals, the instrument may be used either as an 
ammeter, or as a voltmeter of several ranges (see, for example, 
fig. 288). 

Unipivot Instruments 

To increase the sensitivity of the above types of instruments, 
a modification of the support of the moving part was introduced 
by R. W. Paul. The pivot friction was reduced very considerably 
by the use of the one-pivot method of suspension, and at the 
same time all the advantages of the form of double-pivot sus- 
pension were retained, so that a sensitivity corresponding to 
one subdivision per micro-ampere is obtainable for the uni-pivot 

The construction is shown in figs. 289 and 290. A circular 
coil is suspended about a spherical core of soft iron between 
the poles of a permanent magnet. 

Fig. 290 

Fig. 290 shows the detail of the coil support. A vertical 
spindle carries a light counterpoised pointer, and rests on a 
polished jewel at the bottom of a cylindrical hole drilled in the 
soft iron sphere. 

1-'JG. 288 

Fag* 4(10 


The cylindrical spring at the upper end has a very, slight 
lifting effect on the coil, and produces a restoring couple when 
the coil is deflected ; it also serves as a lead for the current to 
the coil.. The current leaves the coil by the flexible wire shown 
at the lower extremity. 

The centre of gravity of the moving part is at the point of 

A simple device is included to raise the point off the jewel 
when the instrument is not in use. This is shown under the 
coil in fig. 289, which gives the general appearance of the instru- 
ment when one pole piece and the magnet are removed. 

For many purposes it is necessary to be sure that the ammeter 
or voltmeter used is reliable to a fair degree of accuracy. The 
instrument may be calibrated in the laboratory (e.g. by potentio- 
meter and standard cell), but it is here suggested that each 
laboratory be provided with one form of ammeter or voltmeter 
(or both) which has been tested at the National Physical 
Laboratory, and is provided with a correction certificate. 
Such instruments should be retained at laboratory standard, 
and the working instruments checked against such standard 


When smaller currents are to be measured, use must be made 
of some form of galvanometer, an instrument which is not as 
robust as the above, it must be levelled before use, and further 
the current must be calculated from the observed deflection 
produced by it. 

Thus it is not as convenient and simple to use as the ammeter, 
but has a sensitivity which is impossible to attain in the latter. 

The increase in sensitivity is largely produced by a more 
sensitive method of suspension. The friction of the pivot is 
entirely removed by the use of a fine suspension of silk or 
phosphor-bronze. The suspension carries a small concave mirror 
by means of which small movements of the moving part may be 
magnified. The two common methods of producing such 
magnification are by use of 

(1) a lamp and scale, 

(2) scale and telescope. 

(1) Lamp and Scale Method 

In this method a beam of light from an incandescent lamp or 
Nerst filament is directed by means of a lens on to the concave 
mirror, which reflects it to a scale some distance away.* The 
greater the distance, D, between mirror and scale, the greater 

*The scale should be placed a distance away from the mirror equal to its 
radius of curvature. The condensing lens, over which is stretched a vertical 
wire, acts as an illuminated object whose image, a circular patch of light with 
vertical black line, is used to measure deflection. 

i _,. ^ 


the magnification produced. When the mirror rotates through 
an angle, 0, the reflected beam moves through twice that angle, 
causing a movement of the spot of light, say, d cms. on the scale. 

Hence tan 20 =■=> 

To measure such deflection it is essential that the mirror 
should produce a clear image. For this reason, with the size 
and character of the mirror available on such a suspension, the 
usual maximum value of D is one metre. 

(2) The Telescope and Scale Method 

A scale is set up horizontally at about one metre from the 
galvanometer mirror, and a telescope, usually under the central 
graduation of the scale, is turned towards the mirror. When the 
mirror, which should be a plane one for this method, is parallel to 
the scale, the latter may be seen in the telescope. The reading 
of the scale in coincidence with the cross-hair in the eyepiece of the 
telescope is noted. When the mirror moves a second scale reading 
will coincide with the cross-hairs. The difference for an angular 
rotation, 0, of the moving system may be accurately measured. 

If this is equal to d cms. then, once more, tan 2 = ^- 

It is advisable, for simplicity of reading, to use either a telescope 
with an erecting lens or prism or to use a scale provided with 
' inverted ' graduations, so that the scale appears the right way up. 

For many reasons it will be apparent that the instrument 
cannot be calibrated once for all. The scale distance is variable , 
the suspension may of necessity be replaced, and so on ; there 
is, therefore, no permanent direct reading scale as in case of 
ammeters and voltmeters. 

The usual method of converting scale readings to the corres- 
ponding currents is to obtain the ' sensitivity ' of the instrument. 

The sensitivity of a galvanometer may be defined in many 
ways. In addition to being a ' reduction factor,' converting 
scale divisions to current, as indicated above, the sensitivity also 
gives an indication of the possibilities of the instrument. How- 
ever for a comparison of two instruments another factor will be 
discussed later. 

Current sensitivity may be defined in either of the two following 
ways : 

(a) The number of mms. deflection produced on a scale one 
metre away by one micro-ampere, i.e. by io- e ampere, or 

( p) the number of micro-amperes required to produce one mm. 
deflection on a scale one metre from the galvanometer mirror. 

The second definition is perhaps from some points of view 


the better, but (a) gives a value which varies directly with the 
property measured. It must be understood that a sensitivity 
of 1000 mms. per micro-ampere means only that a very small 
current produces a deflection which would correspond to 1000 mms. 
deflection when a micro-ampere passes, assuming the deflection 
to be still proportional to the angular movement of the coil. 
It is purely a theoretical mode of expression. The deflections 
measured never actually exceed from 5° to 8°. 

Volt sensitivity may be similarly defined, substituting micro- 
volt for micro-ampere in the above. 

It will be seen that for a fixed current strength the current 
sensitivity is greater for a bigger coil resistance, i.e. a bigger 
number of terms. 

Volt sensitivity, which deals with a fixed potential applied to 
the galvanometer, is the greater the smaller the value of the 
resistance of the coils. 

Determination of Current Sensitivity 

The galvanometer whose resistance, G, has been determined, 

is connected in series with a megohm (10 6 ohms) and a cell of 

known electromotive force, E. The deflection on the scale one 

metre away is noted = d x mms., say. 

E x 10 * 
Now the current passing is-; — micro-amperes; from 

which the number of mms. deflection produced by one micro- 
ampere may be calculated. 
An alternative method is shown in fig. 291. 

Fig. 291 
A steady accumulator whose electromotive force, E volts, 
is known accurately — either by a potentiometer comparison with 
a standard cell or by measurement with a calibrated voltmeter — 
is connected in series with a high resistance, R ohms, through a 
commutator, C, to a low resistance, S, which is in parallel with 
the galvanometer of resistance, G. 


The mean deflection d, say for both positions of the com- 
mutator is obtained. 
The current causing the deflection is readily calculated, for 

the effective resistance of the shunt and galvanometer is u . - ; 

the main circuit current is therefore equal to* 

^Wo W 

where B is the battery resistance which, if an accumulator is 
used, is negligible compared with R. Hence neglecting B in (i), 
the current through the galvanometer is 

E S 

C« = cT — x /o _i_ r\ X io 6 micro-amperes 


E'Sxio 6 

micro-amperes (2) 

R(S + G) + SG 

. . . . d 

Hence the sensitivity is ^- mms. per micro-ampere. 

Thus the sensitivity having been obtained the deflection 
produced in any case may be converted to the corresponding 
current values so long as the suspension remains the same and 
the scale is at the same distance, one metre, from the galvano- 
meter mirror. 


Galvanometers for the measurement of steady direct current 
may be subdivided into two general classes according to the 
nature of the moving part: {a) moving needle, (b) moving coil. 

(a) Moving-needle Galvanometers 

The student will be familiar with the simple tangent galvano- 
meter. In this form a small magnet is suspended at the centre 
of a coil of wire of n turns. When a current of C, expressed in 
electromagnetic units passes through the coil, placed in the 
magnetic meridian, the magnetic field set up causes a deflection, 0, 
such that the restoring couple due to the horizontal component 
of the earth field, H, balances the couple due to the magnetic 
field of the current, and we have : 

C = — tone, (3) 

where r is the mean radius of the coil. 
In general the assumptions upon which the formula is developed 


* N.B. — If S is very small, c.f. R, the factor „ , ^ may be omitted as this 

S + G ES 106 

is a little less than S. If further S is small, c.f. G, (2) becomes j^ s ^tg) 


are not justified, and corrections should be applied to allow for 
the width of the coil, etc. ; such corrections are but seldom used, 
and the galvanometer is not used as a small current measurer. 

A modification, known as the Helmholtz galvanometer, consists 
of two similar coils placed on a common axis at a distance apart 
equal to the radius of either. 

A short magnet is supported at the mid-point between the 
centres of the coils. As with the tangent galvanometer, the 
plane of the coils is placed in the magnetic meridian, then using 
the same notation as before, 

C^-^tanS; (4) 


n is tne number of turns in each coil. 

These two forms are such that the absolute value of the current 
may be simply calculated, but they have no claim to great 

Sensitive Current Detectors 

The result expressed in (3) for a simple tangent galvanometer 
shows that, for this form of instrument, the deflection for a given 
current may be increased by 

(a) decreasing H, the control field, 

(p) decreasing r, 

(y) increasing n. 

There is a limit to methods ((J) and (y) which is prescribed by 
the practical problem presented. 

Further, as r is decreased and n increased, the conditions 
upon which (3) was developed no longer hold. 

(a) may be best carried out by the use of an external control 
magnet which neutralizes the value of the horizontal component 
of the Earth's field. Further, by using an astatic pair of magnets 
or groups of magnets the control effect may be reduced without 
interfering with the magnitude of the deflecting couple. 

The above conditions are embodied in the Thompson (or 
Kelvin) and Broca galvanometers. 

The Thompson Galvanometer 

The Thompson galvanometer is illustrated in fig. 292. The 
control magnets, EE, enable the galvanometer to be used in 
any position independently of the Earth's field, the suspended 
astatic system is shown at the right of the diagram. Each set 
of magnets is placed at the centre of a pair of coils, FF. The 
magnets are mounted on a light rod, the whole being supported 
from a torsion head by an unspun silk or a quartz fibre. 


The magnitude and direction of the control field may be varied 
by alteration of the positions of EE along the vertical rod shown. 

Sets of coils of different resistances are usually supplied. 
Using high-resistance coils a sensitivity of about 600 mms. per 
micro-ampere may be attained, i.e. the instrument will detect 
currents of the order of io~ 9 ampere. 

The movement of the magnet is detected, using the small 
mirror, B. 

Fig. 292 

The Broea Galvanometer 

This instrument is shown in fig. 293. The moving ' astatic 
pair ' is made up of two vertical magnets having consequent 
poles as shown. The control field is due to the magnet, B, which 
moves in a ball socket to any desired position. 

The coils, EE, may be of any suitable resistance, say, 10, 100 
or 1000 ohms as required for sensitivity in the circuit. They 
are connected in series, one at each side of the spaCe occupied 
by the centre magnet, and arranged to produce a field in the 
same direction. 

As in the Thompson galvanometer, an aluminium vane, G, 
moves between two parallel plates which may be adjusted by 
the rods terminating in the metal knobs, CC. The damping 
of the system may be altered by an adjustment of the distance 
between these plates. Currents of the order of io -10 ampere 
may be detected, 


The Thompson and Broca galvanometers are most advan- 
tageously employed as very sensitive current detectors whose 
sensitivity may be rapidly adjusted over a wide range by the 
control magnets. 

Both forms require levelling before using. 

Fig. 293 

(b) Moving-coil Galvanometer 

The moving-coil galvanometer is constructed in a manner very 
similar to the ammeter already described. A phosphor-bronze 
strip, F, fig. 294, acts as suspension to the coil, C, which is free 
to move in the gap between the pole pieces of a permanent 
magnet, one of which is not shown in fig. 294, and a soft iron 
cylinder, I, which is screwed to a brass plate as seen in the 
lower figure. 

In this case the magnetic field in the air gap is approxi- 
mately radial, as shown by the broken lines in the lower figure 
(fig. 294). 

The current enters the coil via the phosphor-bronze strip, F, 
and leaves at the under end by means of a helix of phosphor- 
bronze, S. The control is mainly due to the twist of F in such 
a case. 

If B is the magnetic flux in the air gap, A the area of 
one turn, n the number of turns, and t the restoring couple per 
unit angular twist of the suspension, F, we have, for the case of 


a radial field, when a current, c, passes, causing a deflection, 0, 
BnAc = t6, (5) 

Gc = t6 

G = BnA and is the couple on the 
coil for unit current and is called 
the galvanometer constant, 
or c = kQ, 

when k is a constant and equal to =^-i . 



Fig. 394 

For small deflections, 6 = tan 0, i e. © is proportional to d, the 
deflection on the scale. 

From (5) above it will be seen that to increase the sensitivity, 
t must be made as small as possible, and B, n and A as large 
as possible. 

To decrease t we may decrease the cross-section of the strip 
or increase its length. 

For a circular wire it was shown on page 102 


Ttt 7' 

The most profitable method of decreasing t is therefore to 
reduce the cross-section, i.e. use a fine suspension. This is 
limited by the fact that very fine suspensions are also very 
fragile and somewhat difficult to use. From this point of view 

Fig. 295 

Pagt 469 


phosphor-bronze is the most satisfactory material and is almost 
universally employed . The usual type of moving coil is supported 
by a fibre of the greatest convenient length. 

The increase of B, n, and A must be considered together as the 
terms are interdependent. 

Assuming the magnet to be fully saturated, the value of B 
in the air gap depends on the size of the gap. The smaller the 
air space, the larger is B. The effect of increasing nK is to 
increase the size of the coil, and consequently the air gap, and 
hence decrease B. 

In practice it is usual to use a standard size of air gap which 
allows a frame of such size to be wound with wire, and move 
freely in the available space, that the best compromise between 
these two effects is obtained for a maximum value to the product 

An example of galvanometers of this type is shown in fig. 

Such instruments may have a sensitivity as great as 1500 mms. 
per micro-ampere, i.e. the instrument will measure currents of 
the order io -10 ampere. 


The moving-coil instrument must be adjusted before use. 
The coil is first released, then the instrument is levelled by means 
of the levelling screws, so that the coil does not touch either the 
pole pieces or the iron core, and is thus able to swing freely. 

The instrument may be used in any position, and the coil 
is turned by means of a torsion head, which carries the suspension, 
until the plane of the coil is approximately parallel to the sides 
of the magnet. It is inadvisable to make this adjustment, unless 
the reflected beam does not fall on the scale, for there is a danger 
of breaking the suspension. 

Onwood Moving-coil Instruments 

The Onwood galvanometer differs somewhat from the above 
general types in that it requires no levelling, is not so fragile, 
and is more convenient to move. 

The difference is in the method of suspension as seen in fig. 296. 
NS is the permanent magnet with a soft iron core, d, which is 
drilled to the centre. From this point, h, is a short suspension 
c, which is of sufficient length to support the coil, a, from the 
point, g, clear of the fixed parts. A small mass, e, at the end 
of the rod, /, keeps the system vertical. 


The current is led through the frame to the iron core, and 
thence through the suspension, and the tube, b, to the coil 
which it leaves by means of a flexible ligament at the base of 
the coil, but not shown in the diagram. 

Fig. 296 

Since the coil is supported at the centre it will be equidistant 
from pole pieces and iron core alike for all positions of the 
instrument not too far removed from the horizontal, i.e. the 
instrument does not require levelling. 

A small spring, not shown, at the upper end of the suspension 
protects the latter against sudden shocks. 

The galvanometer is a very small one and the sensitivity 
claimed ranges from 20 to 500 trims, per micro-ampere according 
to construction details. 


To select the galvanometer most suitable for a particular 
experiment many factors must be considered. To decide firstly 
between the two general types described above, their relative 
advantages and disadvantages are discussed below. 

Moving-needle Type 

(1) The value of the control field is affected by the proximity 
of external magnetic fields. The instrument may register a 
deflection when no current circulates through the coils when 


magnets are moved in the laboratory, or dynamos set in motion, 
or even by a passing electric tramway car. Thus there is 
some uncertainty unless such stray fields are eliminated. 

(2) The moving magnet may have its moment altered by the 
field set up when a current passes through the coil. 

(3) Unless the instrument is made ' dead beat ' the needle 
takes a long time to come to rest and is therefore somewhat 
troublesome to use. This may be overcome to some extent by 
using an external damping circuit, consisting of a solenoid 
through which a current may pass when the circuit is closed. 
This solenoid is placed near the case of the instrument and the 
circuit closed momentarily as the needle swings, in such a way 
as to produce a field which opposes its motion. 

(4) The damping of the instrument is independent of the 
resistance of the circuit in which it is placed. 

(5) The needle may be supported by means of a quartz sus- 
pension which has the property of returning after deflection to 
its original position. However, the slight torsion in the fibre 
is usually neglected, and the expression for the current is therefore 
not strictly accurate. 

Moving-coil Type 

(1) The moving-coil galvanometer has a large permanent 
magnet, and is therefore practically unaffected by stray external 

(2) No demagnetizing effect is possible on the moving part. 

(3) The movement of the coil may be very rapidly arrested, 
even when the instrument is not ' dead beat,' by connecting the 
ends of the coil together through a very low resistance as the 
coil passes its rest position. 

(4) The electromagnetic damping varies with the value of 
the resistance of the external circuit. If used in series with a 
low resistance this damping is very high, and the coil may take 
several minutes to attain the full deflection, a fact which is very 
often overlooked when the instrument is used. 

(5) The zero-keeping quality of the suspension depends upon 
the degree of sensitivity attained. The torsion at the fibre is 
taken into account in the expressions for the current. 

Having decided, from the conditions of the experiment, the 
type of instrument most suitable, the next problem is, What is 
the most satisfactory galvanometer resistance ? 

The Order of Resistance of a Moving-coil Galvanometer which is 
most sensitive in a Given Circuit 

In general terms we may state that if the current is to be of 
a fixed value, independently of the galvanometer, considering all 


factors, the sensitivity will be proportional, approximately, to 
VG, i.e. will increase with increasing resistance. 

If a fixed potential difference is to be measured the sensitivity 
will obviously be greater the smaller the resistance ; approxi- 
mately, the sensitivity is proportional to —j^, i.e. high resistance 

i VG 

for detecting current, and low for detecting small potential 

The problem usually presented is, given an external resistance, 
R ohms, what is the best value for G, the galvanometer resistance 
for measuring the current due to a fixed electromotive force. 

We saw (pages 467-8, equation (5)) that the couple due to the 
current c is nABc, and if E is the electromotive force in the circuit 

-ETG (6) 

Further, it was shown on that page that in a galvanometer 
there is but a limited space between the pole pieces to obtain a 
maximum sensitivity. Fig. 297 shows a cross-section of the 

frame which will just move freely in this space. Let the cross- 
section of the whole of the windings be a sq. cms. and assume 
that the windings entirely fill the space with copper. If p be 
the mean perimeter of the coil windings, we have length of wire 

used = np, cross-section of the wire = — 

r n 

Therefore G = — *- = ■ * • 

a a 

where o is the specific resistance of the wire, say, copper, 

i.e. ft = VT (7) 


Hence the couple due to the current 
= BwAc 

T> A 155 _E 

from (6) and (7). 


The condition for the couple to be a maximum and therefore 
produce a maximum effect is that 

Vp^ R + G 
be a maximum. 

We have seen that a and p are constant, so that the condition is 

p , „ is to be a maximum, 

K -f- (j 


or that —7= -f- y^Q be a minimum, 

i.e. R = G. 

That is, under the circumstances stated, the maximum sensitivity 
is obtained when the galvanometer resistance is equal to the total 
eocternal resistance. 

It should be noted that the galvanometer resistance in this 
discussion refers to the copper coil resistance only. The resistance 
of the suspension should be included in the value R. This being 
so, it is apparent that there is a lower limit beyond which the 
resistance of a coil may not be reduced with any advantage. 

If the galvanometer is to be chosen as a detector or measurer 
of small differences of potential the most suitable instrument 
will be one of low resistance (e.g. for thermo-couple work). 

Other qualities of galvanometers to be considered when 
making a selection of galvanometers are : — 

(1) Damping 

The damping of the moving part in a galvanometer apart 
from external artificial agency may be considered due to two 
separate causes. 

(a) The damping due to the viscosity of the air. This is present 
in moving coil and needle alike, and is approximately proportional 
to the angular velocity of the system. It is always present, but 
is usually small. 

(b) Electromagnetic damping. In the case of a moving magnet 
the amount of damping due to this cause is very slight when the 
magnet is in a non-metallic case, e.g. when the coils are wound 
on wood or ebonite. 

This is the reason for the long and troublesome wait which 
occurs before the needle returns to its rest position. This may 
be reduced as described under. However, in either case the 
amount of damping is obviously independent of the external 

In the case of the moving-coil instrument, the suspended coil, 
when closed by an external circuit, is moving in a strong magnetic 


field. Under such circumstances the electromotive force 
induced in the circuit sets up a reverse current in the closed 
circuit, which is therefore brought to rest. 

The value of the damping current depends on the magnitude 
of the external resistance, and may become very great for a low 
series resistance. 

For many purposes it is necessary or convenient to have a 
galvanometer such that the moving part very rapidly returns to 
the zero position after being deflected. A galvanometer having 
this property is said to be ' dead beat.' 

In both types of instrument this may be brought about by 
increasing the electromagnetic damping. In the moving-needle 
type this is accomplished by encasing the needle in a copper, or 
similar metal, case. The movement of the needle sets up eddy 
currents in the copper and the magnet is rapidly brought to rest. 

The moving coil may be rendered dead beat by winding it on 
a metal ' former ' or frame. This constitutes a closed metallic 
circuit, and the desired result is obtained. Alternatively, if the 
galvanometer is wound, for ballistic purposes, on a non-metallic 
frame, e.g. bamboo, the same result is obtained by facing the 
coil with a thin sheet of copper foil (cut into a ' picture frame ' 
which is the same size as the edge of the coil). 

(2) Period and Constancy of the Zero 

A galvanometer which is very sensitive has a long time of 
swing ; and also, due to a very fine suspension, some trouble may 
arise due to the ' creep ' of the zero. However, the type of 
instrument used in experiments in this book will not be of the 
extremely sensitive order at which this trouble arises. The main 
cure for the trouble lies in the selection of suspension, and that 
really involves the selection of an instrument maker who will 
take the trouble to minimize this fault. Beyond this the correc- 
tion of residual effect must be solved by the ingenuity of the 
experimenter as applied to the particular experiment involved. 

Quartz for moving magnets and phosphor-bronze for moving 
coils cause least trouble in this respect. 

For average work 5 to 10 seconds per complete swing will be 
best value for direct steady current measurement or for use 
in ' null ' methods. 


The Ballistic Galvanometer 

A galvanometer suitable for measuring a quantity of electricity 
is called a ballistic galvanometer, and has the following essential 
features : 


(1) The periodic time of swing, T, of the moving part is fairly 

(2) Damping of the moving part is very small. 

The first condition is fulfilled by making the moment of 
inertia of the needle or coil which forms the moving part as large 
as practicable and by reducing the controlling forces, for 

— £ 

where I is the moment of inertia of moving part, and t is the 
restoring couple per unit angular displacement. 

Thus by increasing I and decreasing the restoring forces, T, 
the time for one complete swing, is increased. 

The second factor, damping, is reduced in a way which depends 
on the instrument (needle or coil). 

As seen when considering damping (page 473) the electro- 
magnetic damping only may be reduced. The air damping is 
usually small. 

(3) A third condition is, that when used to measure a quantity 
of electricity, the whole of the transient current shall pass before 
the needle or coil moves from the zero position. Should there 
arise a case in which the quantity of electricity to be measured 
takes longer time to traverse the instrument, due, for example, 
to inductance in the circuit, the time of swing of the needle must 
be increased, by loading it, so that this third condition is fulfilled. 

As indicated above, the galvanometer may be of the moving- 
needle or moving-coil type. We shall develop a relation between 
the throw or angular deflection in either type, and the quantity 
of electricity which passes. 

Moving-needle Type 

This type of ballistic galvanometer consists of a needle sus- 
pended by a fine quartz or unspun silk fibre, at the centre of 
two coils, through which the quantity of electricity, Q, passes. 

The control in this case is either the Earth's field or a control 
magnet. The needle in its zero position is arranged at right 
angles to the axis of the coils, so that when a current passes a 
field is set up at right angles to the control field. 

Let G be the galvanometer constant, i.e. the field due to the 
coils for unit current circulating through them, 

H the value of the control field strength, 

I the moment of inertia of the magnet about the axis 
of suspension, 

M the magnetic moment of the magnet. 


Suppose a current of strength, c, to pass through the coils 
for a very small interval of time. Since the third condition above 
holds, the needle will be at right angles to a field of strength Gc, 
and will experience a turning moment, GcM. 

This couple will produce an angular acceleration in the 
needle. Hence (page 53) 

IB .= GcM, 

16 = GM fcdt = GMQ (8) 

At the end of the swing, the needle having turned through an 
angle O , the kinetic energy of the moving needle, \1Q % , has 
been reduced to zero in doing work against the magnetic force, 
Hra (m being the pole strength of the needle) at each pole. 

The work done is 2Hm f cos O J 

= Kml (1 — cos O ) 

= 2MHsin 2 -^> 

where I is the distance between the poles. 

Thus 2MHsin 2 -^ = -I0 2 , 

2 2 

l0 2 = 4MHsin 2 ^ (9) 

Hence squaring (8) and dividing by (9) 

T Q 2 M 2 G 2 , x 

4HM sin 2 ^ 

We have also for the period, T, of the suspended needle in 
the control field, H, 

t T'MH , ^ 

I= -^- («) 

Equating (10) and (11) : 

Q 2 MG 2 = T 2 MH 

„ . ' O "" 4T5 2 

4H sin 2 -^ ^ 
^ 2 

Hence : 

^ T H • e o 

Q = «'G' sm i •••• < I2 > 


In developing this result we have assumed that the original 
kinetic energy of the needle is wholly used in moving the 
magnet against the field, H, through an angle, O . If, however, 
there are any frictional forces, i.e. the needle is slightly damped, 
some energy will be required to overcome this force, and the 
result will be that the observed angle of swing, say, a, is not 
truly of the magnitude given in the undamped case. The true 
value for Q is therefore dependent not on the observed value, a, 
of the angle, but on some slightly bigger angle O . 

It will be shown that O = a(i +-)> where xis the logarithmic 

decrement of the suspended system, 

o-"-B(-+i)] ™ 

Moving-coil Ballistic Galvanometer 

There are several disadvantages in the moving-needle instru- 
ment as in the case of measurement of steady direct current, 
e.g. (1) and (2), pages 470-1* These defects are overcome in the 
moving-coil type of ballistic galvanometer. 

The electromagnetic damping is reduced by using a coil 
wound on a bamboo frame ; T is increased by using a fine phos- 
phor-bronze strip suspension. 

To obtain the connexion between Q, a quantity of electricity 

discharged through the galvanometer, and O , the first throw, 

we may first of all assume that there is no damping. 

Lei: G be the galvanometer constant, i.e. the couple acting on 

the coil when unit current passes through it. 

t the restoring couple due to the suspension, for unit 

angular displacement. 
I the moment of inertia of the suspended system about 
the axis of suspension. 
As for the moving magnet galvanometer (p. 475) let c be the 
value of the current at a time t, then 

Gc = I0 (13) 


Gfc . it = GQ = 10 (14) 

Considering the kinetic energy of the coil, we have, if is the 
original angular velocity given to the coil by the impulse due to 
the discharge of a quantity of electricity Q, 

K.E = £I0 2 


This energy is used in twisting the suspension through 0. 
At any angular displacement 0, the restoring couple is t0 ; to 
twist through a further angle dd, the work done is t0 . dd ; i.e. 
the total work done in deflecting the coil is 

f. =^ (15) 


Thus we have 

t0 o 2 _ 10* 
2 ~~ 2 

I0 2 = T0 O * (l6) 

Squaring equation (14) and dividing by (16) we have 

I = ^H- (17) 

Again the time of swing of the coil is given by 

T = 2* -yjl- 
or * = 7T 

Substituting this value in 17 

T 2 t = G 2 Q2 
4* 2 ~" t0 o 2 

Q -f £-t (l8) 

or if a is the observed first swing, and x is the logarithmic decre- 

o-H-K'+i) ;•■«*■» 

Suppose now we send a steady current of known magnitude, c, 
through the galvanometer and observe the steady deflection, <p, 

Gc ■-« T9» 

or _ = - 

Hence Q=l.t.*( I+ ?) (i 9 ) 

76 -K <p 2\ 2/ v 

It should be noted that the preceding paragraph gives a very 
ready method of finding the ballistic reduction factor for the 
galvanometer. We see that 

^ , , , T x steady current 

Q = ka where k — — J „ — -, — 

2re X steady deflection 

Subsequent deflections may be converted to quantity by multi- 
plying these deflections by k. 

See also page 481, equation (23a) for the moving magnet type. 

To Correct the observed first swing (a) for damping in either form 
of galvanometer. 

The observed first deflection, a, is reduced by damping forces 
which are proportional to the angular velocity. 

The equation of motion, therefore, must be modified from the 
simple form : 

IB + F0 = o, 

where F may be t or MH, according to the type of instrument, 
to include a term proportional to the angular velocity. Let this 
term be K0, then the equation of motion becomes : 

10 + K0 + F0 =0, 
an equation similar to that already dealt with on page 27. 


The amplitude of the system is therefore O e 3l ', using the 
notation of page 161. 

_K T 

Thus the first observed swing, <x x = 6 Q e *i 4. 

K T ' 

Logarithmic decrement, x = -^ • - , 

2X 2 


a x = O e~\, 


O = a i<^ 



In a ballistic galvanometer K is small and although T may be 


rge, I is also large, and -^ is small, i.e. X 

may be neglected in comparison with unity, 


large, I is also large, and -^ is small, i.e. X 2 and higher terms 

4 1 


i.e. . ».-«(i+i)- 

x may be obtained by one of the methods of pages 162 to 164, 

and the correcting factor fi -f - ) obtained. The galvanometer 

must be in the circuit of the experiment when x is obtained, so that 
the damping is the same. 

Another way of correcting for damping in the moving part 
does not involve an independent measurement of x. 

Suppose that in addition to noting the first swing, a lr we 
observe the next swing on the same side of the zero, a 3 , we have : 

K T _K 3T _k 5T 

ai = o e~*i'*; cc 2 =d e *"*; a 3 =0 e *i"4,-etc; 

or & = ^ = e + § T (20) 

a 2 a 3 

K T ( + * T )-* 

i.e. a. x =0 e-* 4 = © |^ aI ; 

or by (ao) a ' = 9 »t-^' 

ie - *•— (£)' (3I) 

Thus, suppose there is an error of one per cent in the deter- 
mination of an angle, f — J is liable to two per cent error or 

(— ) may be in error by one-half per cent. 

The correction for damping obtained in this way will be 
sufficiently correct for many experiments. 

Of course a x , a 2 , and <x 3 are not measured as angles. The 
corresponding scale deflections, <5 X , <5 a , and <5 3 , are measured. 

Now — = tan 2(Xl for small deflections, 

d 3 tan 2<x 3 

i.e. since under these circumstances the value of tan 2a is 
approximately the same as 2a, 

_£i __ «i. 
<5 3 a 3 

So that (19) for the moving-coil galvanometer becomes 

Q_t.i.«*(«!y (22) 

* re <p 2 \a 3 y 

and (i2«) 


o-Sff-f-®'! <*3) 

In the above method the damping under the conditions of the 
experiment is obtained for the correction factor. 

Equation (23) may be modified to conform with (22). For 
if a steady current, c, produces a deflection q>, 

cG = H tan y, 


Q=7 t w s +Cl)l (23fl) 

To Reduce the Excessive Damping in a Moving-coil Ballistic Galvano- 
meter used in a Low-resistance Circuit 

It was shown in the discussion of damping that the moving- 
coil galvanometer has a large amount of electromagnetic damping 
when in a closed circuit of low resistances. This would make 
the use of a moving-coil ballistic galvanometer inadmissible in 
many experiments unless some means were taken to reduce the 

The quantity of electricity passing through the instrument 
is, in general, due to some current change. If, therefore, the 
galvanometer is in circuit whilst such change takes place, and is 
then allowed to swing freely when it has received the impulse 
due to that cause, the damping is almost eliminated. 

'////A i=i , o 

'///A ^ B 




Fig. 298 

By use of a compound key, such as that shown in fig. 298, 
this may be brought about. The three brass strips mounted on 
an ebonite block (shaded) are connected to separate terminals, 
as is the stop S. The battery circuit is connected to A and S, 
and the galvanometer to C and B. When the key is depressed,' 
C and B, A and S are connected, but B and A remain insulated 
by the ebonite stops shown (shaded). When the key is released 
A and S are broken ; impulse, due to induction or whatever 
cause is operative, is given to the galvanometer, and then C and B 
are disconnected by the upward move of the key. If the time 
interval between the break of the battery and galvanometer 
circuits is small compared with the period, T, of the coil of the 
galvanometer, the latter will swing an amount which is almost 
independent of electromagnetic damping due to the low 
resistance circuit. 



If such a key is used for a moving-coil instrument the damping 
correction used should be obtained from observations of the coil 
when swinging freely in open circuit. 

The Grassot Fluxmeter 

The Grassot Fluxmeter is an instrument which performs the 
same function as a ballistic galvanometer, as for example in 
experiments on pages 426, 557, 582, 357, etc. However, the 
instrument is specially designed for measuring magnetic field 
strengths directly. 

Fig. 299 

It is a suspended coil instrument which depends entirely on 
the electromagnetic damping for control. The coil, D, is 
supported by a single cocoon silk fibre, which has a negligible 
torsional control, from a flat spiral, E (fig. 299), to eliminate the 
effect of shocks. The current enters and leaves the coil by the 
silver strip coils, H, as shown in the figure. The usual iron 
core, B, and permanent magnet pole pieces, N, S, supply a constant 
magnetic field in which the coil swings. 

The light frame, FF, carries a pointer and a concave mirror, 
not shown in the figure. The points, T T ', correspond to terminals 
on the case of the instrument. 

In general use a search coil, C, of known mean area and number 
of turns is connected to T,T'. 

The search coil is placed into the magnetic field to be measured 
and the induced electromotive force causes a current to flow 
in the closed circuit and so produces a deflection which may be 
measured either by a lamp and scale arrangement, using the 

Fie, 300 

Page 483 


concave mirror, or, if sufficiently large, by direct reading of the 
pointer over a scale. 

The instrument therefore produces a deflection which is 
proportional to the total quantity of electricity which passes 
through it. 

Fig. 300 shows the general appearance of the instrument in 
use to measure the magnetic distribution along a magnet. 

The method of calibration of the graduated scale and general 
possibilities of the instrument will be apparent from a con- 
sideration of the theory of the instrument. 

Let R = the resistance of the circuit, i.e. of the search 
coil, C, and leads, and the suspended coil, D, 

L = the self-induction of the whole circuit, 

I = the moment of inertia of the coil, D, about the 
axis of suspension, 

E = the E.M.F. induced in C ] 

c = the current in the circuit I at any instant, 

v> = the angular velocity of the coil J 

K = induced E.M.F. set up in D for unit angular velocity, 

G — the galvanometer constant, i.e.. the couple exerted 
on the coil due to unit current passing through it, 

A = the couple due to air resistance for unit angular 
velocity (see damping, page 473.) 

Expressing Ohm's Law for the circuit, we have : 

cR = E - IT- - Ko>, .(24) 


-i(*- L s-*-> 

The equation of motion of the coil is : 

t d * e r A , ^ 

1 dtf =Gc ~ A< * -to) 

Substituting the above value for c and remembering that 

Ci>= — 


T ^6> _ GE _ LG dc _ /GK \ 
dt R R dt \R +AJ G> - 

Now L, G, R, K, and A are constants ; further, if the coil 
starts from its rest position when no current passes, and becomes 


deflected through an angle, a, by the quantity discharged through 
it, the coil is at rest at the end of the swing when no current 
again passes through it. 

Therefore, integrating the last equation over the whole swing 
with respect to t, we have 


/M,?(!f + A).. 

fEdt = (^ + K) a (26) 

Thus it appears from (26) above that the value of the deflection, 

<x; is determined by / JLdt, since the other terms are constant. 

For example, if the search coil is placed in a magnetic field, 
thereby setting up an induced E.M.F. in the circuit, the total 
deflection is independent of the speed of insertion or withdrawal 
of the search coil. 


Now in general the value of A is small and the term -^~ may 

be neglected in comparison with K. And hence we have : 

Edt= Koc, 





With this approximation the deflection a is independent of 
the value of R, i.e. the search coil may be replaced by another, 
provided that R does not become very large for then the approxi- 
mation is not justifiable. 

For the purpose of this account the noteworthy feature is 

that the deflection is, as for the ballistic galvanometer, proportional 

g • dt, which passes through the 

coils, i.e. a is proportional to the quantity. 

The standard form of fluxmeter has a scale, graduated in 
maxwells ; this is graduated experimentally. The deflections 
corresponding to known flux change define points on the scale, 

and since a is proportional to f Edt, which itself is proportional 

to B, the number of lines cut, the subdivision of the scale between 
such fixed points is a matter of dividing angles into equal parts. 
The experimental arrangements for the instrument in measur- 
ing magnetic flux and quantity of electricity are described in the 
chapters where such measurements find a place. 


For many of the experiments for which the instrument is used 
the graduated scale is too coarse. Use is then made of the 
mirror, using a lamp and scale arrangement as in the ordinary 
galvanometer. The deflection on the scale is proportional to 
the quantity of electricity discharged through the instrument. 
A full scale deflection is produced in the instrument by a definite 
flux change in a search coil ; the position of the reflected spot 
of light is noted, as is the movement of the pointer on the grad- 
uated scale. By repetition, the corresponding values are 
obtained. From such observation the value of the flux change, 
corresponding to 1 mm. scale deflection (at, say, 1 metre), is 
deduced from the graduated scale deflection. 


The preceding account has included the method of measuring 
(1) steady direct current, (2) quantity of electricity. 

We now consider briefly methods available for the measure- 
ment of varying currents. Such currents may be subdivided 
into (a) currents of short duration at regular or irregular intervals, 
(b) alternating currents. 

(a) To detect currents of this kind, and also to measure the 
time interval between such short duration currents use may be 
made of 

The Einthoven Galvanometer 

The principle of this instrument may be understood from 
fig. 301. A 'string/ CC, of fine platinum or tungsten wire is 
supported vertically between the poles of an electromagnet. 
If a current be sent down the string, the latter will be deflected 
m a direction parallel to the face of the pole pieces. In the 
diagram, the direction of movement is shown by the arrow, a, 
for a field in the direction, SN. 

To observe the movement, the pole pieces of the magnet are 
drilled, as shown by the broken lines, in a direction parallel to 
that of the magnetic field. Light from a bright point source is 
concentrated by a condenser, CF, placed in one hole, and the 
movement of the illuminated string is magnified by a telescope, 
DE, placed in the other. 

The " string ' is usually attached to the ends of two small 
springs which keep it stretched. The tension may be altered to 
any desired amount by a micrometer adjustment at one end. 
The string and tension-varying device are supported on a frame 
which may be removed bodily from the gap between the pole 
pieces. * 


The deflection produced for a given current depends upon 
the tension on the string and the strength of the magnetic field. 
The former may be varied as indicated above, whilst the latter 
may be adjusted by regulation of the current flowing through 

Fig. 301 

the coils of the electromagnet. The sensitivity of the instrument 
may be therefore varied over a wide range quite simply. At a 
fixed sensitivity the deflection is proportional to the current. 

The usual method of detecting deflections is to use a photo- 
graphic arrangement. The beam of light emerging from ED if 
allowed to fall on a screen forms a shadow image of the string. 
This is reflected on to a cylindrical lens which forms an image 
of the central portion of the line, i.e. if the string is still and a 
photographic plate is moved vertically at the focus of the lens 
a straight white line is produced on the plate when developed. 
If a current passes, the string moves, and a shift of the shadow 
image results. The point focus of the cylindrical lens is thereby 
deflected. This results in a lateral displacement of the lme 
image on the photographic plate. The magnitude of the displace- 
ment gives a measure of the current strength. The natural 
period of the string is small and it rapidly returns to the rest 
position when the current ceases to flow (not more than a few 
hundredths of a second is required). Thus if a succession of 
small currents pass in the circuit, the instrument detects them, 
even when but a few hundredths of a second interval occurs 
between successive currents, whereas an ordinary galvanometer 
would not distinguish the break between them. 

To measure the interval of time between successive impulses, 
or the duration of one of them, a time scale is imprinted on the 
record of the current by means of a 'time marker' as shown in 
fig. 302.' This consists of a device for intercepting the light at 

Fig. 302 

Page 4&G 


regular intervals and thereby making transverse white lines 
across the photographic plate on which the current is recorded. 

A small motor is made, as shown in the figure with a soft iron 
armature of, say, 10 teeth. Intermittent current is supplied 
to the electromagnets by connecting them in series with the 
circuit of an electrically maintained tuning fork, the impulses 
given to the motor are therefore regular. Suppose the fork 
vibrates 50 times per second, the synchronous motor will rotate 
5 times per second. This drives a circular disc provided with 
projecting arms as shown in the figure. These arms are usually 
allowed to move across the beam of light illuminating the 
apparatus. The figure shows five such spokes, the fifth being 
broader than the others. A spoke intercepts the light 25 times 
per second and makes a time scale on the photographic plate of 
■£ g second. 

By making 20 spokes to the wheel T hs second graduations 
may be obtained. The width of a division may be varied by 
varying the speed of movement of the plate. For a continuous 
record a cinematograph film may be used. 

Of course a number of strings may be used, and the image 
focussed simultaneously on the same film by placing a 45 right- 
angled prism in the path of each image. The whole is therefore 
concentrated on to the slit in the camera box. The arrangement 
of the parts is seen in fig. 303 which shows a plan of the apparatus. 

prisms ~_ 

Fig. 303 

(b) Alternating Current 

Two types of instrument may be used for the measurement 
of alternating current : 

(1) The oscillograph which gives the wave form of the current. 

(2) Ammeters or dynamometers which give the effective current. 

The Duddell Oscillograph 

Fig. 304 shows the essential features of this form of instrument. 
A thin phosphor-bronze strip, ss, is supported over a small ivory 
bobbin, P, and fastened at the lower ends. The tension on the 
strip may be adjusted by regulation of the tension on the spring 
suspension of P. 


■ If a current is passed through the loop, ss, the two strips will 
suffer a deflection in opposite direction, and consequently rotate 
a mirror, M, which is attached to both. If the current direction 
is reversed, the direction of rotation is also reversed. 

Fig. 304 

Thus, for alternating current the mirror would rotate back- 
wards and forwards, provided that the natural period of the loop 
is small compared with that of the alternating supply. 

The image of a source of light reflected by the mirror on to 
a scale or screen by M would therefore be drawn out into a line 
for such an alternating current. 

If the image were focussed on to a photographic plate in a 
camera, and the plate were moved at constant speed in a direction 
normal to the beam of light and the direction of vibration of 
the image, the trace on the plate would be approximately a sine 
curve. The form of the current time curve may then be investi- 
gated from the record. 

To see the wave form on a screen the photographic arrange- 
ment could be dispensed with, and a mirror made to rotate and 
reflect the first beam on to a screen. The linear patch of light 
is again converted to wave form by thus adding a constant 
velocity normal to that produced by M. 

The amplitude of the curve gives an indication of the maximum 
current strength, which is approximately proportional to it. 

This instrument can also be employed for many of the purposes 
to which the Einthoven galvanometer may be used. 

'For a description of the other forms of alternating current 
measurers, attracted iron ammeters and dynamometers, the 
student is referred to any text-book of electrical engineering ; 
e.g., T. F. Wall : " Electrical Engineering." Methuen. 



The Wheatstone Bridge 

The student will be familiar with the Wheatstone net as shown 
in Fig. 305. When the bridge is balanced the relation 


Q = s W 

holds. For maximum sensitivity, using a fixed galvanometer 
and battery and measuring a resistance R, it has been shown* 
that P, Q and S should be chosen so that 

Q 2 = BG, 

P 2 = RG 

R + B 

S 2 = RB 

R + G 

R + G R + B 

If choice of galvanometer is practicable, it should have a 
resistance comparable with the other arms. When P ^= Q =cr R 
^= S maximum sensitivity is obtained when G — P. 

Measurement of the Resistance of a Galvanometer 

This may be done in several ways, of which one or two are 
given below. 

Fig. 305 

Kelvin's Method 

In this method the galvanometer acts as its own detector of 
balance in a Wheatstone net. The galvanometer is placed in 
the arm, DC (fig. 305), and the galvanometer of that figure is 
replaced by a single-way key, so that B and D may be connected 
together when the key is depressed. 

When the battery circuit is completed, a steady current flows 
through the galvanometer. P, Q and R are adjusted until on joining 

* See Gray : "Absolute Measurements," Vol. I, p. 331 [1 



B to D throughthe key, no change is produced in that deflection, 

P R 

when ^- = -p, where G is the resistance of the galvanometer. 

W ** 

The usual difficulty with a sensitive galvanometer is that the 
steady current is too large. The galvanometer may not be 
shunted in this experiment, but the E.M.F. applied may be 
reduced, e.g. the cell may be connected through a high resistance, 
and leads from the end of a small fraction of the resistance may 
be taken to A and C instead of the battery directly applied. 
This, however, decreases the sensitivity of the method. 

For a moving-magnet instrument it is better to apply the cell 
directly, and to reduce the steady deflection to zero by adjustment 
of the control magnet, or, if that is not sufficiently strong, by 
the adjustment of an external bar magnet. The sensitivity is 
thereby retained. 

For a moving-coil instrument no such adjustment is available, 
and the simplest course to follow is to clamp the coil of the instru- 
ment and find its resistance using another galvanometer as 
detector in the usual way. 

See also page 643. 

The Carey Foster Bridge 

The Carey Foster Bridge, shown in fig. 306, is a modification 

of the metre bridge. *lt is provided, as seen, with four gaps, 

PR R 1 Q 

« 4 D D l E E 1 F F 1 

nafe 6 6i g— 3r-arer 

FlG. 306 

CC 1 , DD 1 , EE 1 and FF 1 , which may be closed by the insertion 
of resistances. 



Suppose the gaps be closed with resistance Y in CC 1 , R in DD 1 , 
R 1 in EE 1 , Z in FF 1 , as shown in fig. 307. The battery, E, and gal- 
vanometer, G, are arranged at points, A C and B D, which corre- 
spond to the same points in the theoretical net diagram, fig. 305. 


If the point D is chosen such that no current passes through the 
galvanometer, we have, from (1) : 

R = Y+^+SxP , , 

R 1 Z + r t + (ioo-*,)p' - K) 


P is the resistance per cm. of bridge wire, 

x x the length, SD, 

DT = (100 — x x ) if ST is one metre, 

r x is the value of the resistance at the soldered junction, S, 

r 2 the resistance at T. 

If the simple metre bridge were used to compare R and R 1 , 
i.e. Y = Z = o, we should have a balance at, say, l x cms., such 
that, neglecting r x and r 2 for the moment, 

?L — h? 
R 1 (100 -l x ) P ' 

Now suppose Y = vp and Z = zp, equation (2) becomes : 

?L= (y + *i)p 

R 1 {{z + (ioo-x l )}?' 

if r x and r 2 be neglected. 

It is obvious from a comparison of the two results that the 
Carey Foster bridge functions as though the length of the wire 
were increased, i.e. the same error in obtaining a balance point 
corresponds therefore to a less percentage error in the Carey 
Foster bridge determination. 

Comparison of the British Association Ohm and the Legal Ohm 

The comparison of two resistances, very nearly equal, serves 
to show a common use for this bridge. In the following method 
it will be seen that the end resistance is eliminated. 

The resistances R and R 1 are made approximately equal to 
the values of Y and Z. Connecting as in the fig. 307 with, say, 
Z, a B.A. ohm, and Y, a legal ohm, whilst R and R 1 are each, 
say, 1 legal ohm, we should obtain a balance at a point, D, 
x x cms. from S, so that : 

5. = Y + '1 + *iP ( o\ 

1 Ri Z + r 2 + (100 - x x ) p' u; 

If now Y and Z are interchanged, Z being connected in the 
gap occupied in fig. 307 by Y, and Y replacing Z, a balance for 
such an arrangement could be obtained at a distance x 2 from S, 
from which we have : 

J* _ Z + r x + x iP () 

R 1 Y + r 2 + (100 -*.)?' ■•" w 

Equation (3) may be rewritten : 

R _ Y +?!+*!! 

R + R 1 Y + Z + r x + r % + ioop' 
and (4) similarly becomes : 

R _ Z + r x + x 2 p 

R + R 1 Y + Z + r x + r 2 + loop ' 

Equating numerators of these equations we have : 

Y -f r x + #jP = Z + r x + # 2 P> 

Y-Z = (*.-*,)p (5) 

i.e. the difference between Y and Z is equal to the resistance 
of the bridge wire between the two points of balance. It is 
independent of the value of r x and r 2 , and of the total length of 
the bridge wire. 

To obtain the value of the B.A. ohm in terms of the legal 
ohm, we have, Y being the legal ohm : 

B.A. ohm = {1 — (x t — *j)p} legal ohm. 

The Value of the resistance of the bridge wire between the 
points of balance may be obtained by calibrating the bridge 
wire (see page 495), or, if the wire is uniform, the following 
simple method will serve. 

Resistance of Unit Length of Bridge Wire (p) 

(1) The bridge connexions of the main experiment above re- 
main as before . R and R x are approximately equal, and may very 
well be the same as above. Y and Z are replaced respectively 
by a fraction of an ohm, say, Y 1 , and a stout strip of copper 
of negligible resistance, say Z 1 = o. 

Following the same procedure as before, a balance is obtained 
at a distance x x x from S, when Y 1 and Z 1 are in the positions of 
Y and Z in fig. 307. When Y 1 and Z 1 are interchanged the 
balance will move to another point, x 2 x from S 1 . 

Then by equation (5) : 

Y*-Z 1 = (V-*i 1 )p. 
or since Z 1 = o, and Y is known, 

Xn X-i 

'2 "*1 

Hence, using this value of p, the difference between Y and Z in 
the first case may be evaluated. 

This method may obviously be applied to any similar case, 
and a comparison between two nearly equal resistances obtained. 



Experimental Details 

The resistances and Y, Z, Ri and R 2 are connected by means 
of stout copper strips. These will have practically zero resistance, 
and the small difference to be measured will be truly the difference 
in the resistance of the coils. It is also essential, of course, that 
all connexions be very tightly screwed, for the same reason. 

In performing the second part of the experiment, i.e. to find p, 
two or three sets cf observations should be made ; the exact 
number will depend on the total resistance of the bridge wise. 

For example, Y 1 should be made -i, *2, -3, -4 ohm successively, 
and p calculated in each case, from which a mean value is obtained. 
It may be found that, -4 ohm acting as Y 1 , no balance is obtain- 
able. In that case the total resistance of the bridge wire is less 
than «4 ohm. 

The following set of observations shows the order of the 
result obtained in such an experiment. 

Using Z = 1 B.A. ohm, Y = 1 legal ohm, 
x 2 = 51-2 cms., x x = 48-2 cms. 

The wire was not calibrated. 


Y r in ohms. 





x x x in cms. 




no balance 

x 2 x in cms. 




p ohms per cm. 
_ Y 1 




x% x^ 

Mean value of p = -00377. 
Hence since 

Y - Z = (# 2 - x x ) P , 
Legal ohm — B.A. ohm = (51-2 — 48-2) (-00377), 

B.A. ohm = 1 — 3 x -00377 = -989 legal ohm. 
(2) Another equally simple method for finding the resistance 
per cm. of bridge wire may be used. Suppose Y, Z, R and R 1 
(fig. 307) are all 1 ohm coils (of the same kind, e.g. B.A. ohms). 
R and R 1 are set up as before. To introduce a small difference 
between Y and Z, one of them is shunted with, say, 10 ohms, 
i.e. the net result is ££ ohm. If now the process described above 
is carried out, interchanging the 1 ohm and effective ££ ohm, we 
have two balance positions, l x and l 2 cms., say, and 
Y -Z = {l x -l 2 ) 9 , 


Hence p is determined. 

5- <*i-^ P. 


To Construct a Resistance Coil of Known Magnitude 
To construct, for example, a i ohm coil, a wire with small 
temperature coefficient between io° and 20° C. is selected, say 
manganin, and the resistance per cm. of the specimen available 
is obtained by finding the resistance of about ioo cms. 

The length of wire required to have a resistance not less than 
1*1 ohms is calculated, and cut off. The insulation covering is 
removed from the ends and, using a non-corrosive flux (say, 
resin or ' fluxite '), the two ends are soldered to two stout copper 
wires which are soldered to flat copper forks, A and B (fig. 308). 

Fig. 308 

A and B and the rods are then fastened to opposite sides of a 
wooden bobbin by means of terminals as shown, and the wire 
wrapped, as in the diagram, in a non-inductive manner round 
the bobbin. The middle of the wire T is freed from the silk 
cover. The resistance between A and B is obtained by the 
Carey Foster method. The middle, T, is twisted with pliers, 
cutting out the end loop until, compared with a standard 1 ohm 
coil by the method of page 491, the balance is in the centre of 
the bridge. T is then soldered in position, and the value of this 
copy of the standard ohm when completed is compared with 
the true standard as already described. 


The simple metre bridge, briefly referred to on page 489, also 
the Carey Foster Bridge, the potentiometer and similar instru- 
ments, depend upon measurement to a point of balance on a 
stretched wire. For simplicity it is often assumed that the wire 
is of uniform cross-section, and that its resistance per cm. is 
consequently constant throughout the length. Further, the 
soldered end and the thick copper connecting strip is assumed to 
be of zero resistance. ' 

In a practical measurement it is better to make no such 
assumptions, but to determine the variation due to these causes 
by a preliminary calibration of the bridge. 


To Determine the End Correction of the Bridge Wire 

For this determination the outer gaps of the bridge (CC 1 , FF 1 
in fig. 306) are closed by short clean thick copper strips of 
negligible resistance, and in the inner gaps are two unequal 
resistances, say a 10 and a 1 ohm coil (R and R 1 ). 

A balance is obtained at, say, x x cms. from A. The 10 ohm 
and 1 ohm coils are interchanged, and a second balance is obtained 
at x 2 cms. Now, if the ' end resistance ' at A is equivalent to 
l x cms. of the bridge wire, and if the end resistance at B is equiva- 
lent to l 2 cms. of bridge wire, we have : 

R. _ x x + h 
R 1 100 — x x + /, ' 

and R1 = *■ + '» 

R 100 — x 2 + 1% 
from which l t and l 2 may be calculated. 

Calibration of the Wire 

This may be done in many ways, of which we will consider two. 

The object of these experiments is to find, at different points 
along the wire, lengths having the same resistance, usually equal 
to that of a gauge employed. Knowing the total resistance, S, 
of the wire, the mean value of the resistance of such a length is 
readily calculated, and hence the correction to be applied at 
each segment taken, to reduce to the mean value, may be deter- 
mined. Alternatively, having obtained the lengths which have 
the fixed resistance, the method of analysis used in the calibration 
of a tube may be applied (page 43). Suppose we wish to test 
every 5 cms. of the wire, and this will be quite sufficient for most 
cases, the following methods may be used. 

(1) Carey Foster's Method 

^ We saw on page 492, that if two resistances, R and R 1 , are 
fixed in the inner gaps of the Carey Foster Bridge, and resistances, 
Y and Z, are balanced, and then interchanged and once more 
balanced, that : 

Y — Z = (% 2 — *i)p> 
where x t — x x is the difference in the balance points, and p the 
mean resistance per cm. between these balance points, i.e. 
the difference between the resistance of Y and Z is equal 
to the resistance of the wire included between the balance points, 
the end resistances being eliminated by this method. 

If now we make Y-rZ equal to the approximate resistance of 
5 cms. of the bridge wire, and so arrange R and R 1 that the 


values of x 2 and x t are made in turn to pass along the whole 
wire, the calibration is completed. This is done in the following 

The value of S, the total resistance of the bridge wire, is found 
either by using another measuring device or the same bridge,* 

Knowing S, the mean value of 5 cms. of the bridge wire may 
be found. A length of wire, preferably of the same material as 
the bridge, is then taken ; its resistance is measured and a 
length cut off 2 cms. in excess of that required to be of the same 
resistance as 5 cms. of the bridge wire. 

This is then soldered to two stout copper lugs as shown in 
fig- 309- The two cms. excess being soldered to the lugs. This 
gauge is used as resistance Y. Z is composed of a thick copper 
connecting strip of practically zero resistance (fig. 310). 

B 1 1 




Fig. 309 Fig. 310 

Carrying out the process of balancing Y and Z, inter- 
changing, etc., we have, if x 1 and x 2 are the two values of SD, 
as before, 

Y — Z = resistance of wire between x 2 and x v 

By altering R and R 1 the part of the wire compared with 
Y — Z may be varied. This process is best done by using 
another bridge wire as (R + R 1 ) seen in fig. 310. PQ is the 
second bridge wire connected with thick copper strips to the 
bridge as shown. The galvanometer, G, is joined to B and D 
by means of variable contacts : the gauge is placed in the left- 
hand side and the zero resistance in the right, the point D is 
chosen near the end of the wire and B adjusted until the galvano- 
meter when closed in the circuit shows no deflection. Y and Z 
are interchanged and, leaving B fixed (i.e. R and R 1 are fixed), 
D is adjusted to some position, D', where a balance is again 
obtained. , The resistance of the length, DD ', is then equal to 
the resistance of the gauge. Keeping the contact at P 1 fixed, 

* The value of S may be found using the same bridge. Referring to'fig. 307, 
the gap, Z, is closed with a stout copper strip of negligible resistance. The 
galvanometer is permanently corrected between points S (x=o) and B. R, R 1 
and Y are adjusted for balance, the whole bridge wire making the fourth arm in 

R 1 
a Wheatstone net. When balanced we have S «■ ^r * Y. This gives S with- 
out allowing for the end corrections. 



the gauge and the strip are returned to their original positions 
and B is adjusted to some position, B 1 , where balance is obtained 
(i.e. a new ratio R : R 1 is obtained). Y and Z are interchanged, 
and the contact D' is moved until again a balance is obtained, 
i.e. it is moved a further distance having a resistance equal to 
that of the gauge. 

This method is carried out until the whole wire, SP, is covered 
in approximate 5 cm. steps. The length of wire (x 2 — #1) for 
the different steps are noted for each part of the scale, and 
tabulated as under. 













0- 5 


+ •15 




— 15 








+ •05 

— 0-20 










- -05 



Mean length = 4*95, say. 

The correction is obtained by finding the mean of column (2), 
and finding what must be added or subtracted from the first 
value in column (2) to give the mean value 

For the 5-10 cm. range the correction is the algebraic sum 
of the first 0-5 cm. range, and the 5-10 cm., and so on. 

A correction curve is drawn which will convert any length 
of the wire to the equivalent length of a uniform wire of the 
same length and resistance. 

Thus, if in use in an experiment the bridge balances at 65 cms., 
and from the correction curve the correction, —0*2 say, is obtained, 
if the wire were uniform and of same length and resistance, 
64*8 cms. would be the balance point, and the resistance ratios 
of the two segments of the wire is 648 : 35 '2. 

It will be seen that in this method the interchanges which are 
frequently made must be assumed to cause no variation in 
resistance. The gauge and the straight copper connecting strip 
are thoroughly cleaned and always screwed tightly in the gaps 
to avoid, as far as possible, such changes. ' , 

This method may be used to calibrate wire, PQ, since the 
contact, B, is moved a distance corresponding to equal increments 


of resistance. It would therefore serve well for the calibration 
of a potentiometer wire. 

Direct Calibration by a Potentiometer Method 

The wire, ST, of the bridge is connected, as shown in fig. 311, 
in series with an adjustable resistance, R 2 , and a 2 volt accumu- 
lator which is in good condition and well charged. A similar 
accumulator is connected in series with a second adjustable 
resistance, R x , and a length of, say, 20 cms. of stretched wire 
of the same material and approximate cross-section as the 
bridge wire. G is a high-resistance galvanometer. 


A B 

3 IE 



C D 


1^— ^ 


Fig. 311 

C is soldered to the wire, M?and with A and B in contact with 
ST, say, between the o and 6 cm. readings, and about 5 cms. 
apart, D is adjusted until no deflection is given in the galvano- 
meter. D is then soldered to M. 

It should be noted that convenient resistances for R x and R 2 
are of the order 20 to 50 ohms. The adjustment of these resis- 
tances is made until the currents are about equal. 

Leaving B in contact with ST, reverse the current in M by 
means of the commutator, K, and move A to the other side of 
B; adjust till again no deflection is produced. Repeat this 
process over the length of ST. 

If the latter is uniform, then of course the length AB will 
be the same. Variations of the cross-section will be made 
apparent by the different values of AB. 

Tabulate the length of AB as, before, and deduce the correction 
curve for the wire. 

This method may also be used for any stretched wire, e.g. 
potentiometer, etc. 

* Alternatively, the potentiometer may be made up of two resistance boxes, 
Rj and R/, in place of the wire M and box R x . 




Using the ordinary Post Office Box method of finding resistance 
is a very ready method, but for resistances of a million (10 6 ) ohms 
or more it is unreliable. The ordinary Post Office Box enables 
a magnification of 1000 to 10 only to be obtained, and in the 
adjustable arm 10000 ohms is the maximum resistance available, 
so, for a resistance greater than 10 e ohms a special method is 

Equal limitations for very small resistances make it necessary 
to employ special methods in this case. 

Some of these special methods are described under. 

Low Resistance 

(1) The Direct Deflection Method 

To find the resistance, r, of a low resistance wire, AB, the 
following simple method gives a fair approximation. AB is 
joined in series with a known low resistance, R, and an accumu- 
lator (2 volt), and an adjustable, fairly large resistance, S. 

Fig. 312 

Having by means of S adjusted a small current, c, through 
the circuit, there will be a drop of potential, cr, across AB, and 
cR between the ends of R. If a high-resistance galvanometer is 
used, the current in the main circuit is least disturbed, and the 
resulting current in the galvanometer is proportional to the 
potential cr or cR applied to it. If this causes a small deflection, 
d x cms., using the usual lamp and scale method, cr oc 0, approxi- 
mately, and 0j oc d x approximately. 

The same galvanometer will have deflection i a cms. (angle 0^ 
when connected to the ends of R. Again cR oc 2 and 0„ oc d 2 


B x d x cr 

~e t ~~ d 2 ~ cR 


r = ^-R 
d % 

Hence r = -=*.- R „. (6) 


This method of observing deflections may be used to find the 
specific resistance of copper. 

A length of copper wire is soldered to two terminals about 
i metre apart on a wooden base, and is connected in series with 
•i or «oi ohm, and a constant source of potential such as a steady 
2-volt accumulator, and a resistance box of o to ioo ohms. (It 
is not advisable to have less than 20 to 30 ohms in the circuit.) 

The ends of the wire AB (r) and of the ^ ohm (R) are con- 
nected to a double pole two-way switch (indicated by the broken 
lines in fig. 312), which is connected to a high-resistance galvano- 
meter, G, the deflections of which may be observed by the usual 
lamp and scale method. 

The deflection given when the potential difference between 
the ends of r and R are applied is measured by taking the reading 
of the deflected spot of light in each case. The battery is then 
reversed and the reading is obtained on the other side; half 
the difference in readings giving d x and d 2 . 

S is adjusted and the experiment repeated for 2 or 3 values 
of the current, and the mean value of r is obtained. 

Hence, putting R = t& or y^r in (6), r is evaluated. 
(2) Potentiometer Method 

A steady lead accumulator E x is set up in the potentoimeter 
circuit AB. 

Fig. 312a, 

The two small resistances to be compared, R and r, are 



joined in series with a third variable resistance S, which is 
adjusted so that the current is the maximum compatible with the 
capacity of the lead accumulator E 2 . 

A galvanometer G, is connected as shown, and a double pole 
two-way switch is used. This should be a mercury cu|> key, 
using well amalgamated copper connecting strips. 

A balance is obtained at H when cups 1 and 4, 2 and 5 are 
joined together. A second balance is obtained at H' when 
2 and 4, 5 and 3, are joined together. 

The P.D. between C and D is cR where c is the steady current 

in the circuit CFS ; in the same way the P.D. between D and F 

is cr. So that if AH = l x and AH' = l 2 , 

cR l x R l x 

— = -i or — = -1 
cr l 2 r l 2 

The length l x is obtained, then l 2 , and finally l x is checked ; 

if any difference is found the mean of l x and l x ' is compared with 

1 2, in the usual way. 

(3) The Kelvin Bridge 

To obtain a more accurate comparison of two small resistances 
than may be obtained by the foregoing experiments, the Kelvin 
Bridge method is used. Fig. 313 shows the general arrangement 
of this network. 

r x and r 2 are equal resistances, as are r 8 and r 4 . These are 
arranged as shown, so that the total resistance, D to N = that 
from N to G and resistances in the arm EK = resistance in KF. 

The two resistances to be compared, if in the form of wires, 
are arranged as at AB and BC ; B being a mercury cup or other 
low-resistance junction. 

The two resistances, r x and r s , are connected to AB at points 
D and E, and the second low resistance is connected to r 2 and r x 
at F and G, the latter being an adjustable point. 


The two pairs of equal resistances (about i or 2 ohms each) 
should be made of wire of the same material, or having similar 
temperature coefficient, and should be placed in close proximity 
to ensure a common temperature throughout the experiment. 
Further, when the current from the battery is sent through the 
circuits, its duration should be short to eliminate excessive 
heating in these resistances, and in the specimen AB and in BC ; 
the latter, presumably, will have different coefficients, and the 
comparison might therefore be upset if such temperature variation 
are introduced. 

The galvanometer between K and N should be of low resistance. 

The points E and F should be near B and the point of contact, 
G, is adjusted until on closing the battery and galvanometer key 
no deflection is produced in the latter; then if the resistance 
DE = r, FG = R, we have 

1 = r -l = *J. 
R r t r 4 

Thus it is apparent that r v r* r z , and r 4 are not of necessity 
equal, but the relation 

r 1 = ^ w 

r 2 *4 

must be satisfied. This condition is satisfied in the construction 
of the bridge which, like the Post Office Box, may be used with 
equal ratio arms or with factors 1 : 10, 1 : 100, 10 : 1, 100 : 1, in the 
usual form. 

To establish the relation,^-= - = -, let us put c = the current 

■K ^2 ?3 

in DE (f), Cj the current in EKF (when in the balanced condition 
no current passes through the galvanometer), c t the current through 
Then, since the current c — c t passes along EBF, the current 

in FG (R) is c. 

The above is the balanced condition of the net of conductors, 
i.e. when the potential at K is the same as the potential at N, 
i.e. when 

cr + c x r x = ctf* _ r _z; 
c x r 2 +cR c % r A r 4 
i.e. r A {cr + c^) = r z {c x Yi 4- cR), 

or c(*r 4 - r 3 R) = c x {r# 3 - r x u). 

But the condition satisfied by the bridge, by construction, is that 

£ = £2 or r x u = r 2 r 9 , 
r t r A 

i.e. c(rr A — r s R) = o ; 

and since c is not equal to o : 

rr k = y,R; 


or 4 = ~ = - fr om (7) (8) 

With such an arrangement of resistances as already described 
the comparison of two nearly equal low resistances such as two 
wires, can be made ; then if a, I, and s, represent the radius, length 
and specific resistance of r, and similar terms apply to R, we have : 

7 o /» 2 y o 

/ 1 2 2 = -* ,whence — may be calculated. 

To obtain the specific resistance of a metal in the form of a 
thick wire or rod, we may carry out the above process using a 
copper wire as a second low resistance, and assume s for copper 
(s for drawn copper at i8°C. =178 x io- fl ). 

Alternatively the copper rod may be replaced by *oi ohm (r), 
and the unknown rod placed in BC (R). The length, FG, to 
balance may be found as above ; whence s for the material : 

•01 r t 

sl/a* r % 

When a definite length of wire, or fixed small resistance, is to 
be measured it becomes advisable to have a variable value for r. 
This may be a variable length of copper wire of known cross- 
section for very small resistances (R), or a series of »oi ohm coils, 
and a calibrated wire in series with the coils for larger resistances. 
This latter method is usually used in the ready-made Kelvin 
Bridge now on the market. 

Of course it must be realised that when the values of R, the 
unknown resistance, is very small, the current passing through 
it must be correspondingly large if any appreciable potential is 
to be attained between the ends, and if the measurement is to 
be at all satisfactory, the reason for a sensitive /oze/-resistance 
galvanometer will also be apparent. 

As examples of the types of Kelvin Bridge on the market, 
consider figs. 314 and 315. 

For measuring the specific resistance of a wire of low resistance, 
the special clamping contact arrangement shown at the back 
of fig. 314 is used. For other forms of low resistance, the ter- 
minals X of that figure and fig. 315 are used directly. 

Such an apparatus has a wider range and is more adjustable 
than the simpler form already described but is, in principle, 
identical with the bridge illustrated in fig. 313. To emphasize 
this, fig. 315 has been drawn showing the internal arrangements 
of fig. 314, lettered to agree with fig. 313. 


The ' ratio arms ' may be adjusted as indicated to give various 
ratios, viz. -oi, «i, i, 10, ioo. The coils which make up 
?xt r* ?z, r if are accurately adjusted to ensure the condition of 
(7) above. 

As shown in the figure, r is the unknown resistance, and R 
may be one or more of the four coils each of -02 ohms, together 
with a part, F'G', of the graduated and calibrated wire, G'G". 
The length of the wire is about 450 mms. and each mm. has a 
resistance of '00005 ohm, i.e. 40 cms. of the wire add -02 ohms 
to the value taken from CG' (-04 in the diagram = GG'). The 
wire is graduated in fractions of an ohm. 

A comparison of figs. 315 and 313 will show the general arrange- 

For example, suppose an unknown resistance r were balanced 

w j 1 en — was made — «oi, by -04 ohm from the coils, and 290 
units of the slide wire, as shown approximately in fig. 315, 

t — R f — 1 = (-04 -f 2qo x -00005) * 01 

\ r, / 

= (-04 + -0145) -oi 
= "000545 ohms. 
This instrument will measure resistances of 10 to -ooooi ohm. 
Fig. 317 shows a second form of Kelvin Bridge made from the 
potentiometer illustrated in fig. 338, and a ' double ratio ' box 
(fig. 316). This enables a range of measurement from 1-5 ohms 
to -ooooi ohm, about. 
The ' double ratio * box contains resistances (fig 317), LN, NN', 

N'N" and N"M, which are respectively \, ^, tttt and 




the total resistance of LM. A similar arrangement holds for 

I'm. 314 

Fig. 3 r6 

Pw 504 



the parts of HI, so that the resistances LN : NM = HK : KI 
(1 : 1) or LN' : N'M' = HK' : K'l (= 10 : 1), etc. The scheme 
of connexions is lettered to conform with the letters of fig. 313. 
A comparison with that figure shows the reason for this scheme : 
R, the unknown resistance, is equal to the resistance between 
E and D when the galvanometer is connected to NK as in the 
case taken, or is ^ ED when the galvanometer is connected 
at N'K', etc. 

The coils to which D is tapped are each «i ohm and the resistance 
of the slide wire is *ooi ohm per small division. Thus, using 
N* K* as galvanometer terminals, a resistance of -ooooi may be 



» 6 vd 

C.3 4-C-* T* VO I* 1-4 


Fig. 317 

High-resistance Measurement 

As already stated, a special method is required to determine 
the resistance of greater value than 1 megohm (i.e. 10 6 ohms). 
One of the simplest ways of evaluating resistances of this order 
is the method of substitution. Fig. 291, page 463, shows the 
scheme of connexions. 

A steady accumulator, E, is connected in series with the 
unknown resistance, R, and connected to a commutator, C, 
whence the current may be sent in either direction through a 
galvanometer, G, which is provided with a variable shunt, S. 

Let the steady deflection due to the current be d cms. on a 
scale one metre away (corresponding to a movement of the 
suspended system of 0°). A known adjustable resistance is 
substituted in place of R, and if on adjusting the known resistance 
a steady deflection d cms. is obtained again, the known resistance 
is of the same value as the unknown. If no variable known 
resistances of the order of the unknown resistance are available, 
the galvanometer is shunted so that one-thousandth part of the 
current is allowed to pass through it. When a large known 
resistance (e.g. 10000 to 20000 ohms from two P.O. boxes in series) 
is placed in series with it and the steady accumulator, a deflection 
of the same order as that given when the unknown resistance is 
placed in the circuit may be obtained. 


Under these circumstances, suppose that d x is the deflection 
caused by the battery when in series with the unknown resistance, 
R ohms, and d 2 when in series with r, the known resistance. 

Then if E is the E.M.F. of the accumulator, B the resistance 
of the battery, and G the resistance of the galvanometer (if 
the latter is a moving-coil galvanometer with a radial field), we 
have, where k is the galvanometer constant, 

Mx = B + G + R' 
kd* = 


S . 

(■» SG , I 

S + G' 

. *i |fr + B)(S+G) + SG} S + G 
'•<*■ {(S + G)(B+G + R)} S 

M S + G 
(,+B)(S+G)+SG W s + ^ _ () 
S(B + G + R) G + R w 

when, as is obviously the case, B may be neglected in comparison 
with r or R. 

If further, G is negligible compared with r or R : * 



d % R » 

i.e., if Trnrry part of the current goes through the shunted galvano- 

* S + G 
meter, — ^ — = iooo, 

d x _ iooor 

1,e - T 2 -~R 

The experiment may be carried out, using an adjustable 
known resistance, and shunting the galvanometer with a known 
shunt. The known resistance is adjusted until d 2 = d v i.e. 
equal deflections are obtained. Then, if G is small compared 
with R, 

R = iooor, (10) 

if the shunt is -^. 

Alternatively, if the known resistance is not sufficiently 
adjustable to cause equal deflections, the values of d x and d 2 , 
both of the same order, are noted, then, neglecting G in com- 
parison with R, we have : 

rf, S, + G 

* But according to p. 473 G should be as high as possible to make G as near 
equal to R as possible for maximum sensitivity when the experiment is per- 
formed without the shunt S. 


Again, if the shunt takes ^^ of the current 

R =-r • iooor. 
d x 

In both these cases, if G is not negligible compared with R, 
then the corresponding formulae for the two cases is given by 
equation (9) above. 

Determine by the above methods the value of the resistance 
of a leaky condenser. Another suitable high resistance to be 
measured by these methods is made by taking a sheet of ebonite, 
say 20 cms. by 5 cms. Two holes are drilled about 15 cms. 
apart, and the surface of the ebonite blackened, round the holes, 
with a black lead pencil. Two terminals are screwed down in 
the holes, and a fine black lead pencil line is ruled between them. 
The sheet is covered with a thin protecting second sheet of 
ebonite to prevent any accidental change in the dimensions of 
the line. 

The apparatus should be set up as in fig. 291, and the values 
of R obtained. 

For example, using a thin black lead pencil line, 

&x = i«95 cms., R = ? no shunt ; 

^2 = 9*3° cms., t = 10000 ohms, shunt fa, 
Hence : 

R = -2-2- x 10000 x 1000 = 47-7 x 10 « ohms. 

Using a thicker line on ebonite, i.e. smaller resistance, the 
first method (equation (10)) : 

deflection 10 cms. when R was in circuit, no shunt, 
deflection 10 cms. when 3640 ohms replaced R, shunt fa. 

Thus, R = 3640 x 1000 = 3-64 x 10 6 ohms. 


The value of a resistance of a specimen of wire, in most cases, 
increases with temperature. 

The relation between R , the resistance at o° C, and R f , the 
resistance at another temperature, t° C, is : 

R, = R (i +at + &*), (11) 

where a and p are constants. 

The constant, p, is small, and therefore over small ranges of 
temperature the resistance is practically a linear function of 
the temperature, i.e. 

R t = R (i + «t) (12) 

expresses the relation, for small temperature ranges, between 


resistance and temperature, (t should not exceed ioo° C. for (12) 
to be valid). 

From (12) we have : 

^a ^ R * ~ R °, 

R t 

and a may be called the ' coefficient of increase of resistance with 
temperature ' for the limited range taken. 

To measure such resistance changes and determine the value 
of a, which is practically constant for all pure metals, we may 
make use of the Carey Foster bridge, since these resistance 
changes are small. This method is specially advantageous for 
such a determination of small differences, as has already been 

The theory of the method is almost identical with that already 
given, so that we may deal with the experiment as such. 

Determination of the Resistance of a Wire at Temperature from 0° C. 
to 100° C. 

A length of thin platinum wire of about 1 ohm resistance is 
mounted on a mica frame in the non-inductive manner shown in 
fig. 318. The ends of this wire are soldered to two thick copper 
wires whose resistance is negligible compared with that of the 
platinum. This arrangement is mounted rigidly in a glass tube, 
with mica or rubber supports as shown at M. The free ends of 
the copper wires are soldered to two long leads, and two identical 
leads, cut from the same wire, are joined (D) together. The 
four free ends of the leads are soldered to copper connecting 
strips, shown in fig. 322, P, P 1 , C, and C 1 . These ensure good 
contacts with terminals under which they are fixed. 

The pair, P, P 1 , are the ends of the leads from the platinum 
wire ; C, C 1 are the compensating leads. 

- The inner gaps in the Carey Foster bridge are occupied by 
two equal resistances of the same order as that of the platinum 
wire ; in the case taken these resistances, R t and R 2 , are both 
made 1 ohm (fig. 319). 

P and P 1 are connected in one of the outer gaps and the com- 
pensating leads in series with a resistance box, S, fill the fourth. 

It must be remembered that R 1} R 2 and S should be connected 
to the bridge by means of copper strips, and every source of 
uncertain resistance, such as a bad or dirty connexion, must be 

All the resistances should be ' non-inductive/ for in this 
method it will be seen that the galvanometer is permanently 
connected in the circuit, and the battery takes the position in 
the sliding contact. 



This is essential, for we must balance the resistance of the 
platinum at the temperature which is fixed by the surroundings, 
so that the current should not pass for any appreciable time 
and cause a heating in the spiral. 


Ur y 





Fig. 318 Fig. 319 

S is a resistance box in which are all values from *i to :eo or 
more ohms. 

The function of the compensating leads will be apparent. 
They are in opposite arms to the bridge to PP 1 , and so, since 
the material and length of the leads are identical with those of 
the leads to the platinum, the value of the resistance of the last- 
named leads is eliminated ; and since the two pairs of leads are, 
in the main, side by side, variation in the resistance of this part 
of the circuit is also balanced. 

In obtaining a balance at a point, A, x cms. from the end of 
the wire, the value of the resistance, S, is so adjusted that A is 
near the middle of the wire. Also the balance is obtained by 
having no immediate deflection of the galvanometer when contact 
is made with the bridge wire. If the current continues to flow, 
a change in the resistance of the platinum wire, due to the 
heating by the current, will cause a deflection in the galvanometer. 

Suppose the glass tube containing the platinum wire specimen 
is immersed in a constant temperature bath and a balance is 
obtained at x cms. from the end of the bridge, we have, if R, is 
the resistance of the platinum at this temperature t°, 

Rj ^ R« + r + x ? 
R, r -f S + (100 — #)V 


where r is the resistance of either pair of leads, p is the resistance 
of i cm. of bridge wire. 

Since R x = R 2 = i ohm, 

R t -f r + x 9 = S +r + (ioo -*)p, 

R t = S + (ioo — 2X) P (13) 

Hence, if p be known, R f may be calculated. 

The value of R is obtained by surrounding the platinum, etc., 
with melting ice. After twenty minutes or half an hour the 
whole of the immersed tube will have attained the temperature 
of o°C. ; the balance is obtained near the centre of the bridge by 
making S have a value S . A is x cms. from the end. If 
after five minutes the balance is still at x , we may safely assume 
that the resistance is at o° C. 

The value of P may be obtained at this stage, by varying 
S to S ' and finding the new balance at x '. S ' is again ad- 
justed to a third value, S ", and the balance is now at x ". 
The platinum wire is meanwhile at o° C. and therefore its 
resistance remains R„, 
i.e. R = S + (100 - 2*0) p, 

Ro = S ' + (IOO "2* ')p, 
R = S " + (I00-2« *) P . 

These taken in pairs yield three values of p, and further give 
by any one, preferably the value of S which makes the balance 
in the neighbourhood of the centre of the wire, a value of R . 

The glass tube and contents are then placed in a hypsometer, 
and a balance is obtained once more, when the temperature of 
the wire has acquired that of the steam which is made to pass 
round it in the hypsometer. 

The value of the resistance at, say, 40 C, 6o° C, 8o° C. about, 
is obtained in the same way by immersing the container in a 
water bath maintained at steady temperatures near these points. 
In each case time is allowed for the platinum to acquire the 
temperature of the bath. 

The values of R obtained are plotted against temperature. 
The result will be approximately a straight line. From the slope 
of the straight line drawn through the observed points, calculate 
a as defined above. 

In the theoretical account of the balance given above it was 
assumed that there were no end corrections to the bridge wire. 

If l x and l 2 are the equivalent lengths obtained in the bridge 
calibration (see page 495), equation (13) would become 
R t + r + x? + h? = S + r + (100 - x) 9 + l 2 ?, 

' R t = S + (100 - 2*)p + (h - h)?- 

Since l x and l t are known the effect produced may be allowed 
in this way throughout the determinations. 




The values of x and (100 — x) should be corrected from a 
calibration curve as obtained in the manner given on pages 
The Platinum Resistance Thermometer. Callendar-Grifflths Bridge 

The platinum resistance thermometer is a temperature measur- 
ing instrument which depends for its action on the variation of 
the resistance of a wire with temperature, as investigated in the 
last experiment. 

The temperature of any enclosure is calculated from the 
observed value of the resistance of a calibrated specimen of 
platinum wire. 

The instrument consists of a length of fine platinum wire, 
wound in a non-inductive manner on a mica frame. The ends 
of this thin wire are soldered to two thick leads which are con- 
nected at the other ends to two terminals on the cap of the 
porcelain tube which contains them. 

Arranged parallel to these thick leads are a second pair, of 
identical material and size, which are likewise soldered to two 
more terminals on the cap. The other ends of this second pair 
are joined together as shown in fig. 320. 

The terminals are usually marked in a distinc- 
tive manner, P, P, C, C ; C, C being the ends of 
the compensating leads. 

We have already seen (page 507) that the re- 
sistance, R f , of a given specimen of wire, at a 
temperature t° C, may be expressed in terms of 
the resistance at o° C, R , and two constants a 
and in the following manner : 

R, = R (i + <*t + ftf 2 ) (n) 

Thus, if the resistance of the platinum wire 
were measured at o° and two other known 
temperatures, we should have two equations 
from which a and p could be calculated, and 
the wire would be standardized, so that if R t were 
measured at an unknown temperature the latter 
could be calculated from equation (11) above ; 
or having R , a, and p, a calibration curve could 
be drawn showing the relation between R and t, 
and hence the temperature corresponding to any 
resistance could be obtained. 

Alternatively we may define a new scale of 
temperature which expresses the relationship 
rm. 3*u between temperature so defined and the value of 

the resistance in a very simple manner. 


Let R and R 100 be the resistance of the platinum wire at 
o° C. and ioo° C. ; R 100 — R being the increase in resistance for 
ioo° rise in temperature, and is called the ' fundamental interval.' 

The platinum scale makes the size of the degree such that 
each degree rise in temperature on this scale corresponds to an 
equal increase of resistance of the specimen of wire and is equal to 

Rioo Rq 

The platinum scale so defined coincides therefore with the 
gas scale at o° and ioo°, but will differ at other points 
since equation (n) above expresses the true relation between 
R and t on the gas scale. 

Let R be the resistance of the wire at any temperature, t° C, 
or t P on the platinum scale ; by definition : 

tp ==ioo p ~ p • ( x 4) 

■•Moo ss -o 

The difference between t p and t has been found to be given by 


t - U = k- 


100 j 100 

where k = 1-5 for pure platinum, but its value for any particular 
specimen may be obtained by obtaining the resistance R , R 100 , 
and R t at o° C, ioo° C. and a third standard temperature, say, 
the boiling point of sulphur corresponding to a known t° C. 

The values of R , R 100 , etc., may most conveniently be deter- 
mined by using the Callendar-Grimths Bridge. 

The Callendar-Grifflths Bridge 

The Callendar-Grimths bridge is a compact form of the Carey 
Foster arrangement described previously in considering the 
variations of resistance with temperature. Fig. 321 shows the 
general appearance of a modern example of this type of bridge. 

Reference to fig. 322 may make the principle of construction 

R x and R 2 are equal resistances. EF an adjustable resistance 
capable of giving 5, 10, 20, 40, 80, 160, 320, 640, 1280 arbitrary 
units of resistance. ML is a straight stretched wire chosen by 
reason of its uniformity ; T a second parallel wire of the same 
material, which may be connected to ML by means of a short 
length of the same material wire. Thus, the possibility of 
thermo-electrical effect when the galvanometer is connected to 
the wire is eliminated. 

Fig. 321 

Pag* 512 



Suppose the thermometer be balanced at a temperature t p , 
against S units from EF, when the contact is made at O, the 
centre of ML, then using the same notation as before (page 509). 

R t + r + Resistance LO = S -f r + Resistance OM. 
or R, = S. 

If now at a second slightly different temperature t\ the 


Fig. 322 

thermometer has resistance R v and is balanced by moving the 
contact 1 unit of length to the right, 

Rj+rH- resistance of (OL— 1) =S+r-f resistance of (OM+i), 
i-e. R 2 =S -f-2 resistance of 1 cm. of LM. 

If the resistance per cm. of ML is half an arbitrary unit, we 
have : 

Rx = S + 1 arbitrary units. 

The scale on which measurements of MO, etc., are made is 
usually inscribed in arbitrary units equivalent to the movement 
of the contact position. 

If the balance is to the right of the mid-point of LM, the 
number of units must be added to S ; if to the left the number is 
taken from S. 

The length of the wire, LM, is usually sufficient to allow about 
15 units each side of the centre. 

To arrange this simple relation between the length of the 
wire and the arbitrary units, a coil, s, of suitable resistance is 
shunted across the bridge wire. The coil is adjusted to give the 
desired value of resistance per cm. to the bridge wire. 

The arbitrary unit of resistance often chosen is deduced from 
the value of the fundamental interval of the thermometer (i.e. 
the change in resistance for a change in temperature from o° 
to ioo° C). 



The platinum thermometer is constructed with such a resistance 
that its fundamental interval is I ohm and one hundredth part 
of this, i.e. *oi ohm, is taken as the unit. The coils in S have, 
therefore, resistances of -05, *i, etc. 

The resistance per cm. of bridge wire in the example taken 
would therefore be £(-oi) or -005, the effect of 1 cm. change in 
balance being 2 x -005 = •01, i.e. a change of balance of 1 cm. 
corresponds to a change in resistance of the platinum wire of 
one unit. 

The form of Callendar-Grimths bridge illustrated in fig. 321, 
is provided with a scale for the slide wire bridge which is gradu- 
ated from o to 15 units, i.e. the readings are continuous from 
one end of the wire to the other. The balance points give directly 
the number of units to be added to S to give the equivalent 
resistance of the thermometer. The half, MO, of the wire is 
obviously in this case within the instrument in the form of a coil. 

A more recent feature of the bridge is the use of mercury cup 
contacts instead of the usual plug contacts in the adjustable 
arm, S. Fig. 323 shows an enlarged view of one of the contacts. 

Fig. 323 

The coil is connected to two mercury cups inside a tightly 
fitting cover. 

A plug, D, when inserted in the hole, G, corresponding to this 
resistance, strikes a thin circular sheet, E, which is covered with 
baize and is ordinarily held tightly against the hole by means of 
the spring, A, and a spiral spring, F, thus keeping dirt and dust 
from the mercury. When D is allowed to depress this arrange- 
ment, the copper connecting strip, B, which is amalgamated at 
the ends, dips into the cups and the resistance is thereby cut out. 



The total value of the resistance, S, is therefore the sum of the 
numbers opposite the holes without plugs. 

The makers suggest that when not in use all plugs should be 
in to maintain the amalgamation of the connecting strip, B. 

The balance point on the bridge wire may be maintained by 
a rough movement of the slider, followed by a fine adjustment 
by means of the small lever attachment. The position of the 
oalance is read on a vernier which enables one-tenth of the 
small scale division to be measured, i.e. if graduated in lengths 
corresponding to a unit and subdivided into tenths, one may 
read to y^ of the unit. 

Reference to fig. 324 may make clear the internal wiring of 
the bridge. The lettering in this diagram corresponds to that 
of fig. 322. A two-volt accumulator or Daniel cell is connected 
to BB 1 and, say, a Broca galvanometer to GG 1 . PP 1 and CC 1 
are gaps for the thermometer and compensating leads. It will 
be noticed that Rj and R a are contained in the bridge. 

Fig. 324 

The point A in fig. 324 is capable of slight adjustment in many 
forms of the bridge ; so that, if in error, Rj may be made 
equal to R 2 . 

Before using the bridge with the thermometer it is advisable 
to check some of the points of construction which have already 
been described. 

(a) See that the zero of the scale is truly the mid-point ot the 
bridge. This is readily done by inserting equal resistances, e.g. 
two thick copper strips, in the gap provided for the thermo- 
meter and compensating leads. With S = o, find the balance 
point on the wire. This is the mid-point and should coincide 
with the zero graduation ; if this is not so, then probably the 
values of R and R 1 may not be truly the same ; this may be 
further checked by using two practically equal resistances of 


5 or 10 ohms in the gaps PP 1 and CC 1 . A balance is obtained. 
The coils are then interchanged and a second balance obtained. 
The mean of the two readings should be the zero of the scale. 
This method cannot be very well used in the second form of 
bridge wire described above. 

However, for the purpose of comparing resistances the slight 
error in equality will not affect the comparison. 

(b) Verify the relation between arbitrary units of resistance in 
S and the resistance per unit length of bridge wire, and see that 
the resistances in S are consistent within themselves. 

The gap CC 1 is closed by means of a copper connecting strip, 
and an external resistance or resistances are connected by copper 
strips to the gap PP 1 . 

The coil 1280 is unplugged and balanced against an external 
resistance of, say, i2«9 ohms. Suppose x x be the reading on 
the bridge at balance. 

Replace the 1280 coil by the rest of the coils, i.e. 640 + 
320 + .... 10 + 5, and again balance these coils (nominal 
value 1275 units) at a point, x x x , using the same external re- 
sistance. Let l x be the resistance corresponding to a change in 
balance x x x — x x . 

Then coil 1280 — sum of the rest =l x (15) 

Carry out this test using the 640 coil : balance against an 
external resistance making the point of balance at x 2 , say. Cut 
out the 640 coil and, again leaving the external resistance the 
same, balance the rest of the coils (320 -f 160 + 80 + 40 + 20 + 
10 + 5) at a second point x 2 x cms. Let the resistance corres- 
ponding to the change x 2 x — x 2 = h> 
then : 

Resistance (coil 640 — the rest) — l 2 (16) 

Carry out this test with each coil in turn. 
Finally : 

Resistance (coil 5 — o) = / 9 . 

Nine equations are obtained in this way for the bridge described. 
The difference between the first and second equations, where 
each term represents the corresponding resistance is : 

coil 1280 — coil (640 + 320 ... +5) — {coil 640 — 
coil (320 + . . . +5)} =li—h> 
i.e. coil 1280 — 2 coil 640 = l x — 1 2 

.. coil 1280 , l 2 — l x . v 

or coil 640 = + 2 (17) 

In the same way, taking the third equation from the second : 

.. coil 640 , l z — 1 2 ,_™ 

coil 320 = — — — + ± — — » (1-0) 



and so on. Substituting in (18) the value of coil 640 in terms 
of 1280 from equation (17) : 

coil 1280 . l z — h 
coil 320 = • + - 

l z — 1 2 



This process is carried on throughout the range. We may 
express each coil in terms of the largest one. Equation (17) 
and (19) give the corrections to be applied to each coil to make 
them consistent with the largest. 

For a perfect set of coils l x = / 2 = • • • = h = h = tne 
length of wire having 5 units of resistance. 

It is important to note that the unit often taken is -oi ohm, 




x units re- 




640 +320 +...+5 


320 +160 +...+5 



80+40+. ..+5 

80 (+1280) 

40 (+1286) 

20 ( + I28o) 

10+5 (+1280) 

10 (+1280) 
5 (+1280) 

5 (+1280) 


so that for safety the 1280 coil should be out as a permanent 
addition when testing the 80, 40, 20, 10 and 5 unit coils. 

The results may be tabulated as shown above, the last 
column giving the correction to each coil. 

The above calibration gives the mean value per unit length 
of bridge wire. 


For some purposes, and especially when the bridge has had 
considerable use, it may be necessary to calibrate the bridge 
wire. This may be done by a slight modification of the method 
given on page 498. But a new bridge wire should be uniform 
to within 0-3 per cent, and therefore will give readings correct 
to the second decimal place in degrees on the platinum scale. 

Calibration and Use of the Platinum Thermometer 

Having tested the bridge and calibrated the coil and wire as 
indicated above, the thermometer is placed in melting ice and 
allowed to remain there until the resistance remains constant. 
The value of this resistance is noted. 

The thermometer is next placed in a hypsometer, in which 
water is boiled, and again the steady resistance, at the tempera- 
ture of the steam, is obtained (after being in the steam for, say, 
20 to 30 minutes). 

The temperature of the steam corresponding to the atmospheric 
pressure is determined from tables (= b° C. say) then, if R ft is 
the resistance obtained in the steam, and R in ice at o° C, 

R& — Rp 
is the change in resistance per degree centigrade. 

The thermometer is next immersed in the vapour from boiling 
sulphur (see below for method), and from the balance of the 
bridge, when steady, the value of R, the resistance at the tempera- 
ture of the sulphur vapour, is obtained. 

The boiling point of the sulphur in degrees on the platinum 
scale is from (14) : 

_ b(R - R ) 
tp ~ K 6 - R * 

Now the value of the boiling point of the sulphur at 76 cms. 
of mercury pressure is 4447 C, and the value at any other 
pressure, j^mms., is : 

2=4447 + -0904(/> — 760) — «oooo52(£ — 760) 2 . 

The value of t, the boiling point at the pressure which obtains 
during the experiment is calculated and hence, substituting 
t p and t in (14a), the value of k may be calculated. 

It will be found to be very near 1-5 in most cases. The thermo- 
meter is now fully calibrated and may be used, for example, 
to find the B.P. of aniline first in degrees platinum from which, by 
(14a), the corresponding value in degrees centigrade may be 

To Obtain the Boiling Point of Sulphur 

The apparatus of fig. 325 is convenient. This is readily made 
by taking a length of iron tube about two inches diameter, and 



about four or five inches longer than the thermometer. The 
lower end of the tube is closed by brazing on a circular disc of 

The lower end of the tube is covered completely with asbestos 
paper and then wound with nichrome wire of suitable gauge and 
length to produce sufficient heating. For example, the one 

Fig. 325 
shown in the figure was made by winding with nichrome wire 
of -092 cm. diameter and six metres in length. This was 
satisfactory for use with a series resistance on 100-volt mains. 
The length of the same gauge wire may be varied to suit the 
potential of the mains available. It might be found advisable, 
in addition to the thick layer of asbestos around the nichrome 
winding, to surround the boiler with dry sand. The sulphur 
vapour condenses on the upper walls of the tube and runs back. 

The boiling point under the pressure which obtains during 
the experiment, may be obtained from the formula given on 
page 518. 

The Variation of the Resistance 0! a Bismuth Spiral in a Magnetic 

In this experiment a bismuth spiral such as that shown in 
fig. 326 is placed in a variable magnetic field and the resistance 

Fig. 326 

is determined for each value of the field strength. The measure- 
ments involved are therefore (1) the strength of the field, (2) the 
resistance of the spiral. 

The magnetic field should be of as large a range as possible, and 
may be obtained by using a large du Bois magnet as shown in 


fig. 327. Current from the electricity mains is passed through 
a series of frame resistances, R, to the magnet, and an incandes- 
cent lamp, L, in parallel. The current passing through the 
windings of the magnet may be measured on a calibrated ammeter, 
A, and may be reversed by the commutator, C. The ammeter 
is placed at as large a distance from the magnet as is convenient. 


Fig. 327 

By varying the frame resistance, the current passing may be 
of values from, say, to 5 amperes. 

The bismuth spiral is placed centrally between the poles of 
the magnet with the plane of the spiral normal to the magnetic 
field of the magnet and coinciding with the meridian. The 
resistance of the spiral is measured either by a Post Office Box, 
or by the Carey Foster method. In the latter case the scheme 
of connexions shown in fig. 319 may be used, where the gap, 
PP 1 , is closed by the spiral and the resistance, S, may be of any 
value from o to 50 ohms. R x and R 2 may be given convenient 
values ; for a spiral of resistance 15 to 20 ohms make R!=R 2 =20 
ohms, say. 

In either case the determination of the resistance, R 1} say, of 
the spiral will present no difficulties. 

Its magnitude is determined for values of the field corresponding 
to currents in the electromagnet windings ranging from o to 5 
amperes, by, say, '3 ampere steps. 

The initial value of the resistance will depend upon the past 
history of the specimen, but will increase with increasing magnetic 
fields. (The order of this increase will be 10 per cent of 
the original value, depending upon the field change which a 
current variation of o to 5 amperes creates.) 


The values of the current and resistance are tabulated. 
The next part of the experiment is an estimation of the mag- 
netic field corresponding to the currents used. 

This may be carried out in either of the following ways : 

(1) Fluxmeter Determination of the Field Strength (H) 

The fluxmeter is set up, away from the electromagnet, and a 
search coil of the same area as the bismuth spiral is made. 
This is connected to the fluxmeter and introduced between the 
poles of the magnet. The deflection of the suspended system 
is noted, either directly or on a scale one metre away (for the 
weaker field this method is essential). The scale readings are 
converted to maxwells\>y comparison with the graduated scale 
on the instrument as described on page 485. 

If a is the area of the search coil (per turn) and there are 
n turns, the total flux recorded in maxwells (m maxwells, say) is 

m = nail, or H = — gausses. 

na ° 

As explained in dealing with the theory of the fluxmeter 
(page 482) the total deflection is independent of the speed of 
insertion of the search coil. 

The field is found in this way for each of the current values 
used in the resistance determination. 

(2) Ballistic Galvanometer Method of Finding H 

In this method a similar search coil is made and connected 
to a ballistic galvanometer. This forms a low-resistance circuit, 
so that a moving-needle type would be least damped by such a 
low-resistance circuit. This, however, should be removed to a 
very great distance from the magnet and ammeter for obvious 

It will generally be better to use a moving-coil instrument, 
also removed from the magnet, though not necessarily as far 
away as the needle type. In such a case a key such as described 
on page 481 should be used so that the circuit may be broken 
immediately after the passage of the discharge. 

Under such circumstances, when the current is passed through 
the windings of the electromagnet, the search coil being in the 
gap between the pole pieces of the magnet, replacing the bismuth 
coil, a transient E.M.F. is set up in the galvanometer circuit 
through which a quantity of electricity, Q, will be discharged. 
The quantity, Q, as shown earlier, may be expressed by equation 
(19), page 479. 

* n <p 2 \a 3 / 



T is the undamped periodic time (when the key described 

is used), 
a x is the first deflection, 
a 3 is the second deflection on the same side, 
c is a small steady current, 
<P is the corresponding steady deflection. 

The correction for damping (—V will only be applicable 

for the moving-coil instrument when the circuit is broken, as 
described, as soon as the impulse is given to the galvanometer. 

The value of <p corresponding to a current, c, may be obtained 
by passing a current from a 2 volt accumulator through one 
megohm, and the galvanometer shunted by a small resistance. 
The deflection is observed and the current, c, calculated, as on 
page 465. T, a lf a 3 are observed in the usual way. 

Now suppose that the field strength to be measured is H 
gausses, and let c 1 be the current passing through the coil at any 
instant, t, after the circuit is made. During a small interval of 
time, it, let ^N be the number of lines threading the circuit, then 

dE = -=- (numerically), 

, rfN 
or re 1 = -jt, 


where r is the total resistance of the galvanometer circuit (galvano- 
meter, leads and search coil), 



fredt = fdN, 

when a is the area of one turn (= itR 2 ), and n is the number of 

turns, in the search coil, 

~ Han 
i.e. Q = — * 

it (Tc 
i.e. H 

;N(-) 4 - 

[jc tp 2an) 

The bracket term is constant during the experiment and may 
be evaluated and used as a reduction factor throughout. 

The observations described above under distinct headings may 
be carried out successively. The search coil is placed in the gap 
between the pole pieces ; a lf a 3 are observed. The bismuth 

* Alternatively, if the search coil is placed in the gap between the magnets 

2 anil 
and the current reversed in the windings, Q = — — • 


spiral is then introduced and its resistance measured. The 
search coil is replaced and the current cut on. The values, 
"fci 1 * as 1 * are again noted. A new value of the current strength 
is obtained by an adjustment of the frame resistance, and the 
current switched on. The throws are observed and then the 
bismuth spiral resistance is again measured. The check value 
for the throws is again obtained when the current is switched off, 
and so on. 

With either method of working compile a table showing the 
relation between the magnetic field strength and the resistance, 
and plot a curve showing this relation. 

The form of the curve will suggest that the relation between 
R and H is of the form : 

R H = R + d¥L + 6H 8 , 

where R H is the resistance in a field of strength, H, 
R the resistance in zero field, 
a, b are constants. 

By substituting values for two points, find a and b. 

Calculate the expected value of R for another magnetic field, 
in the range of the experiment, as far removed from the first 
two points as convenient. The observed value will be found in 
close agreement with this value. 

Determination of the Absolute Resistance of a Metal Rod. (Lorentz's 

This method of measuring resistance in absolute units is one 
in which a steady drop of potential at the ends of a rod (as 
determined by Ohm's Law) is balanced against one set up by 
induction. The measurements are reduced to those of length 
and time. 

Fig. 328 shows the details ; the sketch shows a copper disc, 
mounted on a horizontal axis about which it may be rotated, 
either by hand or by a motor. A metal brush, C, makes contact 
at the rim of the disc, and a second contact to the disc is made 
at the axle. 

At an adjustable distance, d, from the plane of the disc is a 
coil, D, of N turns of wire, of about the same radius as that of 
the disc (a cms). 

By regulating the position of the movable base, S, the distance, 
d, may be made any length within the limits of the length of 
the base. Fig. 328 also shows, in a diagrammatic form, the 
connexions used in the experiment. AB is the metal rod, say, 
of brass. A single accumulator is connected in series with the 
coil, D, and the resistance, AB. 


If R is the resistance of the rod in absolute E.M. units, the drop 
in potential at the ends of AB is cR, when c is the current in like 

The rod is also connected, by copper wires, in series with the 
disc and a galvanometer. 

Fig. 328 

When the current from the accumulator, E, circulates through 
the coil, a definite number of lines of magnetic force cut the 
disc. If M is the coefficient of mutual inductance of the disc 
and the coil, there is a flux Mc lines through the disc. If, there- 
fore, the latter is made to rotate, an E.M.F. will be set up in the 
disc circuit, which will depend on the direction of rotation of the 
disc for its direction. 

The direction of rotation is chosen such that the E.M.F. set 
up by induction opposes the E.M.F., Re, due to the steady 
current of the coil circuit. By adjusting the speed of rotation, 
or by altering the distance, d (usually by both these methods), 
we may balance the potential due to the two causes. Suppose 
there are n revolutions per second when this balance occurs, as 
shown by no deflection in the sensitive galvanometer G. 

In the disc circuit we have an E.M.F. due to the induction 
equal to 

since Mc lines are cut by any radius for one revolution, 
there are n revolutions per second. is the number of lines 
cut per second, and opposing this is an E.M.F., Re. For balance 

Men = Re, 
or R = nM. 

The resistance, R, is therefore determined if M, the coefficient 
of mutual induction, is known, and n the number of revolutions 
per second is counted. 

Maxwell's formula for the coefficient of mutual inductance in 
the case of two circuits of one turn each, of radius, a, and separated 
by a distance, d, is : 

= 4*«(log.^( I+i ^)-( 2+ -g 5 ) 



The coefficient, M, above is thus Nw, 
1 - e - R = wwN, 

or R= 4 ^N« J 2-303 -10^.(1+^) _( 2 ■ ^ 

16a 2 , , 
The resistance in ohms, where the ohm is defined as io 9 E.M.U. 
is, of course, R x 10-'. 

Experimental Details 

To avoid the complication of the Earth's field in the induction 
of the E.M.F. in the disc, the plane of latter is turned so that 
it is in the magnetic meridian. *" 

The Broca, or other sensitive moving-magnet galvanometer, 
works very well in this experiment. 

If no motor is available, a fair result is obtained by rotating 
the disc by hand. The speed of rotation which may be best 
maintained steady is found by a trial experiment, and the coil 
and disc placed at a convenient distance, d. When all is ready 
the disc is rotated, until the spot of light from the galvanometer 
is brought back to the zero reading. At this stage the rotation 
is maintained constant. The number of revolutions made is 
counted and timed, with a stop-clock, over as long a period as 
the light spot may be kept at zero. 

From this a knowledge of n may be obtained. 

d and a are measured in cms., and the value of R calculated. 

If the length and cross-section of the rod are measurable, the 
value of the specific resistance may be also calculated. 

The following is an experimental result for a determination 
made with such an apparatus. 

400 revolutions of the disc in 172-5 seconds, 
.*. n — 2-325 revolutions per second, 
a = n-8 cms., 
d = 5-0 cms., 
N = 100 cms., 

R = 100 x 2-325 x 148-1 1 2-303 x 1-2761 ( 1 -j- — ^~ ) 

( \ 2227-8 / 

V ^ 2227-8/ ( 
= 35372 absolute E.M. units. 
Specific Resistance : 
Length of rod = 167 cms., area of cross-section = ^(.64)2, 

Specific Resistance = 35372* 1-20, 


= 2732 • E.M. units 
(or -000002732 ohm). 


Resistance of a Battery. (Mance's Method) 

The resistance of a battery may be found by a modification of 
the Wheatstone net as shown in fig. 329. The resistances, 
P,Q,R, and the battery of resistance, B, form the network. 
The connecting wire and key, K, take the place of the battery 
in the usual form of the net. 


Fig. 329 

It will be seen that for all values of P, Q, R, a current will 
pass through the galvanometer, G (i.e. in circuit, DCB). 
The resistances, P, Q and R are adjusted until, on depressing K, 
no change is produced in the steady deflection ; under such 
circumstances we have : 

P R „ OR 

— = — or B = — 
Q B P 

The main difficulty in such a determination is that the steady 
current in the galvanometer causes a deflection which is too 
large to keep the reflected beam of light on the scale ; or, if on 
the scale, the galvanometer is insensitive in detecting change in 
deflections, as seen by the fact that over a large range of resistance 
in the arms, R or Q, the above condition seems equally well 



This is best overcome by placing a condenser of, say, one-third 
microfarad in series with the galvanometer (shown in broken 
lines in the figure). 

The steady current in the galvanometer is therefore eliminated. 
When the key, K, is closed and the bridge is unbalanced, a 
throw of the galvanometer will be recorded. When balanced, 
no kick will be given by the galvanometer when K is closed. 

Another method which may be employed to overcome the 
steady deflection, using a moving magnet galvanometer, is to 
alter the control field in such a way as to reduce the deflection 
to zero, and proceed as first described. This method, if possible 
with the control magnet on the instrument, gives good results 
as the galvanometer is sensitive to the variations produced by 
closing K, when in the zero position. 

If the control magnet is too weak to restore the moving magnet 
to the original zero position, it may be restored by an external 

Using a Kelvin (Thompson) galvanometer, the same result 
may be brought about by sending the current through one pair 
of coils only. The second pair of coils are connected to an 
independent circuit composed of an adjustable resistance and 
cell, and the current sent in such a direction as to produce a 
field in the opposite direction. The magnitude of this current 
is adjusted until the spot of light is at the zero position. The 
effect of tapping the key, K, will then be seen in the resulting 
deflection. The control magnet, in such a case, may be adjusted 
to give good sensitivity to the instrument. 

Conductivity of Salt Solutions (Electrolytes) 

When a direct current is passed through an electrolyte, the 
resulting polarization causes an increase in the resistance of the 
electrolyte. Further, for continued passage of current through 
such a liquid, the resulting decomposition also causes a change 
in the resistance, due to alteration of the concentration of the 

The usual special methods adopted to measure such resistances 
are designed to overcome these difficulties, and are either a 
potentiometer method, or one making use of an alternating 

The second is the one most generally employed. If the 
alternating current is small, and the area of the electrodes large, 
the polarization effect is reduced to a negligible amount. The 
more completely is this brought about when very rapidly alter- 
nating current is used. 

Thus, whereas it is impossible to obtain reliable values for 


the resistance of salt solutions by the Wheatstone bridge method 
in the ordinary way, by using alternating current in conjunction 
with a Wheatstone net, the value of the resistance of the electro- 
lyte may be found. Of course in such an arrangement an ordinary 
galvanometer is useless as detector ; a telephone replaces it in 
the usual modification. 

It is not advisable to introduce any inductance or capacity 
into the net, so for this reason the wire bridge is preferable to 
the post office box type, as the one or ten metres of wire have 
less self-inductance than the coils of the post office box. 
The scheme of connexions is shown in fig. 330. 

AB is a stretched wire of one or ten metres in length, 

R an adjustable resistance, 

S a vessel containing the solution, 

T a telephone, 

C a small induction coil. \ 

Fig. 330 

The alternating current is supplied by a small induction coil 
which will give but small potential. The induction coil is 
driven by a cell and the secondary is connected to the ends, A 
and B, of the bridge. If possible an induction coil without an 
iron core should be used ; the coil with the greater number of 
turns should be used as primary, and the current supplied to it 
should be reversed by a rotating commutator driven by a small 

The telephone, T, should be a head-piece receiver. 

J The vessel, S, shown in fig. 330, should be surrounded by a 

water bath, and be provided with platinum electrodes. The 

temperature of the water jacket is maintained constant, 

as the resistance of the solution varies rapidly with temperature. 

The platinum electrodes are coated with finely divided platinum 
to increase their effective area, and to decrease the back electro- 
motive force due to polarization. If this has not already been 


done, the electrode should be immersed in a solution of 
platinum chloride* and a smaller current passed first in one 
direction, and then the reverse. To prevent an absorption of 
salt from the solution, the platinized electrodes are then raised 
to dull red heat. 

The solution of known concentration is placed in S, and R is 
adjusted so that a point P is obtained near the centre of the 
bridge, such that the sound in the telephone is entirely cut out 
or, as more often happens, until the sound is reduced to a 
minimum, when the usual Wheatstone result may be applied, 

S ~PB' 

It is usual to express the results in terms of the specific con- 
ductivity of the solution, i.e. the reciprocal of the resistance of 
one centimetre, one square cm. in cross-section. 

If r is the resistance in ohms between the electrodes, 
s the specific conductivity, 

then sec— , 



or s = — , 


where A is a constant depending on the dimensions of the vessel 


A may be obtained by finding r for a liquid of known specific 
conductivity. We may take potassium chloride as a standardizing 
solution, using ihe data in the table shown on page 530, 
where the specific equivalent conductivity is the quotient of 
the specific conductivity and the number of gramme molecules 
of the salt per litre. 

It will be seen that for dilute solution this quantity is nearly 
independent of the concentration. 

Make up a solution of KG containing a definite number of 
gramme molecules per litre, and find the resistance, r, of the 
solution when filling the vessel, S. Dilute this solution so that 
the solution contains, say, half this number of gramme molecules. 
The value of the specific conductivity for these two strengths 
may be taken from the above table, and the mean value of A 

Then, using this calibrated vessel, find the resistance of several 
solutions of NaCl from a concentration of, say, 29*25 grammes 

* The Platinum chloride solution to use is made by taking 
1 part platinum chloride, 
30 parts water, 
'008 part lead acetate. 



per litre P^\ to -2925 grammes per litre, by diluting the concen- 
trated solution first made. 

Plot a curve showing the relation between specific conductivity 
and concentration, and specific equivalent conductivity and 

The variation of the resistance of one of the solutions with 
temperature may also be investigated by heating the water baths 
surrounding the cell, S, and the value of the temperature co- 
efficient may be calculated. 

Using a container for the solution, of more measurable dimen- 











AT l8°C. 




0-125 x 10- 3 

125 xio- 8 



1-206 „ 




11-300 „ 




22-00 „ 




50-60 „ 




99-10 „ 










sions, the specific conductivity, and hence equivalent conductivity 
may be calculated in the way usually employed in solids. For 
example, in fig. 331, a uniform tube connects two bottles in 


X j 

Fig. 331 

each pf which is immersed an electrode ; the column of liquid 
conveying the current is practically one coinciding with the 
tube, i.e. of length I cms. and cross-section equal to that of 
the internal cross-section of the tube. The resistance is 
measured as above ; the specific resistance is then directly 
calculated in the usual way, and the experiment proceeds as 
already described. 


Standard Cells 

'the two standard cells in common use are the Weston or 
Cadmium cell and the Clark cell ; of these, the former is better for 
general service. 

The two cells have each a constant electromotive force at one 
temperature, and also a definite temperature coefficient. 

The Weston cell is shown in diagram form in fig. 332. 

Two tubes are arranged as shown, each being provided with 
an external lead which is in contact with the bottom layers. 
These layers consist of pure mercury, M, and an amalgam of 
pure mercury and cadmium, A, respectively. Above the pure 
mercury is a layer of a paste of mercurous sulphate, P, made as 
described later (shown by horizontal shading). Above this and 
the cadmium amalgam is a layer in each tube of pure cadmium 
sulphate crystals, CC. Finally, a layer of a saturated solution 
of pure cadmium sulphate occupies the upper parts of the tubes. 
To make a Cadmium Cell 

The following is a method of construction for the Weston cell ; 
for a more permanent and exact cell the specification given in 
the Report of the British Association Meeting, 1905, on page 98, 
by F. E. Smith, should be consulted. 



(a) Mercury is at first obtained in as pure condition as possible 
commercially. It is then passed through a dilute solution of 
nitric acid, drop by drop. To bring this about take a tube of 
about three-quarters to one inch in diameter, and about 60 to 100 
cms." long ; draw out the end to a smaller diameter and bend 
this smaller tube so as to leave a short U-tube at the end ; the 
shorter end is bent over as seen in fig. 333. 

Fig. 333 

A second length of the wide tube is drawn out at one end to 
a fine capillary of such diameter, that when the tube is filled 
with mercury the latter will just emerge as very small drops. 
This tube is inserted at the upper end of the first, which is filled 
with dilute nitric acid. The mercury is collected in the U-tube 
and passes over into a collecting vessel. 

The process is repeated and the partially purified mercury 
is next distilled under reduced pressure, air being bubbled 
through it during this process. To carry out this distillation 

Fig. 334 
the mercury is placed in a round-bottomed flask, provided with 
a side tube which leads through a condenser to a second round- 
bottomed flask, itself connected to a good water filter pump, 
as shown in fig. 334. 


Into the first flask, through a tightly fitting cork, a narrow- 
glass tube passes under the mercury. A clip regulates the 
inflow of air. When the pump has reduced the pressure inside 
the system, the mercury is heated on a sand bath, and the 
condensed mercury is collected in the second flask. 

The air is allowed to enter through the mercury as a slow 
succession of bubbles. This oxidizes such impurities as zinc 
and minimizes the risk of their distillation with the mercury. 

(b) The amalgam is made by dissolving pure cadmium in pure 
inercury, to make a 12 to 13 per cent cadmium amalgam (i.e. about 
in the proportion of 1 gramme of cadmium to 7 grammes of 
pure mercury). 

(c) The paste is made in a mortar by grinding together pure 
mercurous sulphate, cadmium sulphate and purified mercury in 
the proportion of 8:4:1. The mixture is made into a thin 
paste by the addition of a solution of cadmium sulphate. 

Before filling the cell the platinum wires must be amalgamated. 
In the form of tube shown in fig. 332, this is done by the electro- 
lysis of mercuric nitrate in the glass tubes, using the wires as 
electrodes and reversing the current. 

Another form of glass container for the cell consists of a 
double tube with open upper ends. Corks are selected to fit 
the open ends tightly, and through holes in the corks a narrow 
glass tube may be inserted in each side tube, to carry the wire 
through the cell content to the mercury or amalgam. 

Such platinum wires could well be amalgamated by heating 
to red heat and dipping in mercury. 

Having amalgamated the platinum wires, the tubes are next 
carefully filled as shown in fig. 332. If the open-tube type is 
used, the corks should be finally coated with marine glue or a 
mixture of beeswax and resin. 

The E.M.F. of the cell so formed will be found constant with 
constant temperature. On no account must a current of any 
appreciable magnitude be taken from the cell. 

The International Conference on Electrical Units and Standards, 
1908, adopted the following formula as giving most accurately 
the E.M.F. of the cell : 

E t = 1-0184 — 4'o6 x 10- 5 (t — 20) — 9-5 x 10- ' (t — 20) 2 

+ I0- 8 (tf-20) 3 VOlt, . . « M ... (I) 

where t is expressed in degrees centigrade. 
The temperature coefficient is therefore small. 

The Clark Cell 

This is made in a manner identical with that described above, 
With the exception that cadmium is replaced in this case 


by zinc ; cadmium sulphate by zinc sulphate, etc. Proceeding 
as above, again using pure salts and mercury, the standard cell 
so constructed has an electromotive force expressed by the 
formula : 

E t = 1-4328 - 1-19 x 10- 3 (t - 15) — 7 x 10- 6 (*-i5) 2 . .(2) 

Thus this cell has a larger temperature coefficient than the 

Weston, a fact which explains the more general use of the former. 
When using either form of cell in potentiometer work the 

device of fig. 337 (i.e. a large series resistance) is a useful one as a 

safeguard against damage to the cell when the ' balance point ' 

is not approximately found. 

Comparison of Electromotive Force 

The potentiometer method of comparing two electromotive 
forces is the most satisfactory one. It is assumed that the reader 
is familiar with the direct comparison of two E.M.F.'s, using a 
stretched wire potentiometer. 


In that method a steady accumulator is arranged as at E in 
fig- 335 > and E x and E 2 whose E.M.F.'s are to be compared 
are in turn placed in series with the galvanometer, and a point 
of balance obtained at l x and Z,*cms. respectively ; then, if the 
wire is of uniform resistance per cm., 

E 2 l 2 

Using one of the methods given on page 495, tt seq., the wire, 
AB, could be calibrated and then, more accurately, if l x x and Z 2 * 
are the corrected lengths corresponding to l x and l a 

For an absolute value of the E.M.F. of either cell a third 
balance could be obtained at, say, l 3 for a cadmium cell whose 
E.M.F. is known at the temperature of the experiment and is 

* l t is the mean of, say, three observations, and l 2 the mean of two observations, 
taken alternately ; this eliminates the effect of the variation of E.M.F. of E. 



approximately 1-0184 volts. Whence E x and E 2 may be obtained 
in volts. 

For making such a comparison, the accuracy of the deter- 
mination depends on the accuracy of obtaining the balance 
poiftt. If instead of using a one-metre potentiometer, a wire of 
ten metres (i.e. ten wires in series, each one metre long), 
be used, then each cm. of wire has a potential drop equal to 
one-tenth the drop in the simpler potentiometer, i.e. a movement 
of 1 mm. in the single wire bridge would correspond to 1 cm. 
movement in the ten-wire instrument ; hence by using the 
ten-meter potentiometer the true balance point may be more 
nearly estimated. 

It is often convenient to make such a potentiometer direct 
reading. To do this we arrange that between the ends of a 
definite length there is a fixed potential difference, say, 10- s volt. 

Using a ten-metre potentiometer the most convenient length 
to employ to correspond to io _a volt is 5 mms. This is brought 
about as follows : 

A standard cadmium cell (E.M.F. = 1-0184, assuming tempera- 
ture is 20 C.) is connected to A,* and the jockey makes contact 
with the wire at 1018-4 units of length from the end. 

We have chosen the unit for this purpose as 5 mms. i.e. P is 
fixed at 509-2 cms. from A as in fig. 336. R is now adjusted 

E 1 

. — I— 1 — B 

S.C. G 

Fig. 336 

until the galvanometer gives no deflection, i.e. the current in 
the wire AB, due to E, is such as to cause io -3 volt drop per 
5 mm. of the wire (assumed uniform). Leaving R fixed, any 
other E.M.F. may be found by balancing on the potentiometer 
at, say, / cms. from A or 2/ x 5 mms., i.e. 2.1 x 10- 3 volts is the 
value of the balanced E.M.F. The above assumes that the 
accumulator E remains steady. This should be checked at 
intervals, if a series of comparisons are to be made, by reinserting 
the standard cell, SC, and adjusting R to bring a balance at 
509-2 cms. 

* A resistance of about io 4 ohms should be in series with S.C. until balance is 
almost complete. See fig. 337 and page 537. 


For many purposes it is convenient to replace the wire by- 
variable standard resistances when comparing potentials. A 
suitable arrangement of apparatus using such a method is seen 
in fig. 337. R 2 and R 2 are two resistance boxes, each having a 

E x l |U? K 2 

Fig. 337 

resistance up to, say, 10,000 ohms (two post office boxes do 
very well). 

E is a steady accumulator which is connected in series with 
R x and R 2 . The cells to be compared are connected to A and P 
through a galvanometer, G. 

For direct comparison the resistance (Rj + R 2 ) is kept constant, 
say 10,000 ohms, and R x and R 2 are varied until no deflection 
is obtained when K 2 is closed. The value of R x is noted. 

The process is repeated with the second cell, say, a cadmium 
cell, a balance being obtained for a resistance R^ in the box 
between AP. 

Then, as before ~r = ^j> 

Ji 2 Kx 1 

E R 

for the drop in potential between AP due to E is = ^ — 5 i 

R-i + J^2 + -fc> 
where B is the resistance of the cell E. When a balance is 



E. = 

R x + R 2 + B 
R1 1 


R 2 i + R, 1 + B 
Rj 4. R 2 == Rji -j- R 2 i = 10,000 ohms. 


Rx 1 ' 

Another way of using the above form of potentiometer is 
similar to the direct reading method of using the wire potentio- 


One of the cells to be balanced against E is a cadmium cell. 
If the temperature of the experiment is not far removed from 
20 C, the electromotive force of such a cell is 1-0184. 

R x is given the value 1018-4 ohms (a fraction ohm box may 
be included in series with R a ) . Knowing the approximate electro- 
motive force of the cell, E, the value of R 2 may be estimated 
such that the potential drop in AP ^ 1-0184 volts. The standard 
cell is placed in series with the galvanometer G ; K 3 being open, 
r, of about 10 4 ohms is in series with SC, to avoid damaging it 
during the preliminary balancing. R 2 is now adjusted until no 
deflection is noted in G. K s is closed and the final balance 
verified. The total value of R x + R 2 under these conditions is 
noted and maintained constant throughout the comparison. 

Now 1018-4 ohms have a drop of potential of 1-0184 volts, 
i.e. each ohm corresponds to a potential drop of io~ 3 volt. 

Therefore, when a second cell (E 1 ) is introduced, if R t has a 
new value, R x * ohms at balance (R^ + R 2 X being equal to 
R, and R 2 as obtained in the first test), E 1 is R x x x io~ 3 volts. 

The above methods of making the potentiometer direct reading 
are only suggested for those cases where several comparisons 
are to be made, for under such circumstances subsequent 
calculation is eliminated. 

The experimental arrangements described above are most 
sensitive for comparison of electromotive forces of the order of 
one volt. If now a small difference of potential is to be deter- 
mined it will be apparent that these arrangements are not 
sufficiently sensitive. In the case of a stretched wire, the 
sensitivity increases with increase in length ; therefore, to 
measure a potential difference of the order of, say, io~* volts 
with fair accuracy the potentiometer wire would require an 
extension of several metres of wire, or, what is more convenient, 
the inclusion in the circuit of a resistance several times that of 
the wire (r). For example, if a potentiometer wire of one metre 
were of one ohm resistance, and the accumulator E had an 
E.M.F. of 2 volts, a potential difference of the order of 10 milli- 
volts would balance at a distance of 5 mm. from the end of the 
wire. This could very easily be estimated at 4 mm., or 20 per 
cent in error. Further, under such circumstances the calibration 
of the wire would be of greater importance. 

If on the other hand 99 ohms were placed in series with the 
wire, the drop of potential in the wire would be 1/100 of 2 volts, 
and the- true balance point would therefore be at 50 cms. An 
error in estimating the balance point of 1 mm. would only be 
•2 per cent. 

An example of the application of this method is seen in fig. 
343, page 542. 


Direct Reading Potentiometers 

■> As described above, the potentiometer may be made ' direct 
reading.' Several forms of potentiometer are now on the market 
which are constructed on the above principles and are calibrated 
directly in potential. 

Of the simpler forms of compact manufactured instruments, 
which measures potential of the order of a volt to one millivolt, 
we will describe the form illustrated in fig. 33,8. The internal 
wiring is indicated by white lines drawn on the case, and the 
instrument is an application of the form shown in fig. 336. The 
2 volt accumulator which is connected to the terminals EF, 
sends a current through an adjustable rheostat, a series of coils 
provided with tappings to the studs shown, and through the slide 
wire. This constitutes the main circuit. 

When the current is adjusted to the correct amount, the fall 
in potential along each of the resistances, connected to the studs, 
is *i volt, and along the slide wire, «I2 volt. 

The wire is divided into 120 parts, each corresponding in 
adjustment to a fall of potential of 10- * volts. 

To adjust the current to this strength a standard cell is inserted 
in the gap marked ' potential,' and a galvanometer of low resis- 
tance in the gap marked ' galv.' The two adjustable contacts 
are set at points which correspond to the true E.M.F. of the 
cell at the temperature of the room, e.g. if the E.M.F. of the cell 
is 1-018, the sliding contact at the back of the instrument is set 
at i«o and the contact on the wire made at •018. 

The rheostat is then adjusted so that when the key is closed 
no current is indicated in the galvanometer. 

If the rheostat is not sufficiently large to bring this about 
the battery should be connected to E and G, not EF. This 
introduces more resistance in the circuit as indicated in the 
figure, and, using a normal 2-volt accumulator, the balance for 
the standard cell will be attained. 

To obtain the value of an unknown electromotive force, it 
is connected to replace the standard cell. Leaving the rheostat 
in the balanced position, the sliding contacts are adjusted until 
a balance is obtained, as indicated by no current in the galvano- 
meter. The value of the electromotive force is then directly 
obtained on the calibrated scales. 

The width of the small divisions is sufficiently large to allow 
eye estimation to £ of a division, but is not of suitable range 
for the measurement of the corresponding £ of a millivolt. Thus, 
whereas the instrument would measure a potential of 1-0182 
with a good degree of accuracy, it cannot be used to measure 
•0002 volt with any certainty, nor is it reasonable to expect 

PtlRfi 5 3 B 

Fig. 339 

Pg£B 5J9 


such measurement with an instrument having a range o to 1*5 

For such measurements, as, say, for a thermo-junction, an 
instrument having a range of to 50 millivolts is more suitable. 

Potentiometers of such a range are manufactured by many 
firms, e.g. Nalder, Cambridge and Paul, Crompton, Gambrell, etc. 

Figs. 339 and 340 show the general appearance and internal 
arrangements of such an instrument. 

The current to main circuit is supplied by a 2 volt accumulator, 
B. As seen, this circuit consists of adjustable resistances, Rj 
and R 2 , fixed resistance E and F, MVC, D x and D 2 , and the two 
stretched wires SS and W. 

The range of the instrument is 30 millivolts (o to 30, or 30 to 60, 
or 60 to 90) . The resistance MVC is made up of 29 similar coils of 
such a resistance that when the current is adjusted as described 
below, the potential difference between the ends of each coil is 1 
millivolt. Also the wire, VV, is of a resistance such that for this 
adjustment the drop of potential along its length is 1*2 milli- 
volts. By subdivision, a value of '005 millivolts may be 

Further, the value of the resistance, F, and the slide wire, SS, 
is such that the graduation along SS gives the potential between 
M and N when adjusted. 

To standardize the potentiometer, the standard cell, SC, is 
connected to the galvanometer, G, by means of the double pole 
switch. The point, N, is chosen equal to the potential of the 
standard cell, R x and R 2 are then adjusted until the low resistance 
galvanometer shows no deflection, i.e. the drop of potential along 
MN corresponds to the graduation value. Under such circum- 
stances the potential difference per coil of MVC is one millivolt, 

To measure an unknown electromotive force between .0 to 30 
millivolts, plugs are inserted at the points shown in fig. 340, and 
the unknown potential connected to X. The commutator is 
thrown over so that XX are connected to the galvanometer, 
and a balance is obtained by varying the point of contact, P, 
on MVC and Q on VV. If D 2 is of zero resistance the unknown 
potential corresponds to that between P and Q and is therefore 
obtained directly from the scale. 

Suppose a bigger potential, say, 30 to 60 is to be measured, D a 
is now made of such resistance that there is a potential drop 
equal to 30 millivolts along its length, and therefore for the 

* A modification of this form of instrument, having a range suitable for 
thermo-junction work, is now to be obtained. 


balanced position the potential is 30 + the readings of MVC 
and VV. 

Similarly for 60 to 90 millivolts. D 2 is increased so that for the 
steady current in the main circuit the potential difference between 
the ends of D 9 = 60 millivolts. 


Fig. 340 

To maintain the current in the main circuit at a fixed value, 
the resistance of D x must be decreased by the same amount as 
the increase in resistance of D 2 . This is done in the manner 
shown in fig. 340. 

Thermo-Electrieity— Thermo-Junetions 

When a circuit is composed of two dissimilar metals and the 
junctions of these metals are maintained at different temperatures 
an E.M.F. is set up in the circuit. This electromotive force 
varies with the difference in temperature between the junctions, 
and when one junction is maintained at o° C. is given by 

E f = at + U\ 
where a and b are constants and t expresses the temperature of 
the hot junction in degrees centigrade. 

The direction of the electromotive force depends on the metals. 

It is customary to express E t for any metal with respect to a 
standard metal which is taken as one of the pair. The usual 
choice of standard metal is lead. 

In drawing the curve giving the relation between the E.M.F. 
and temperature, the E.M.F. is taken as positive when the 
current tends to flow from lead to the metal at the hot junction. 



Thus fig. 341 shows the form of these curves for Pb/Fe and Pb/Cu. 
At a temperature t° C. AB represents the E.M.F. developed in 
a Pb/Cu junction and, according to the above rule, the electro- 
motive force is from the lead to the metal at the hot junction. 
Similarly, AC is the magnitude of the electromotive force 
developed from lead to iron at the hot junction. 




1 ^r 


f B 



L^ A 




Fig. 341 

The law of intermediate metals may now be applied to deter- 
mine the value of the E.M.F. developed at a copper-iron junction 
at a temperature t° C, the other junction being maintained at 
o°C, for, according to this law the E.M.F. developed at the 
temperature t°, for a Cu/Fe junction, is expressed by 
E.M.F. Cu/Fe = E.M.F. (Cu/Pb + Pb/Fe) 
= - AB + AC 
. , = + BC, 

i.e. there is an electro-motive force of magnitude BC from copper 
to iron at the hot junction since the BC is positive. 

Reference to fig. 342 will show this from another point of view. 
If junctions, X and Y, are maintained at temperature, t°, Z being 
maintained at o° C, the arrows show the direction of the E.M.F. 's 


of magnitude AC at X counter-clockwise and AB at Y clock- 
wise — a net result BC counter-clockwise. This, according to 
the Law of Intermediate Metals, is the value of the E.M.F. if 
XY are brought together, the lead being removed. The result 
is an E.M.F., BC, in counter-clockwise direction, i.e. from the 
copper to the iron at the hot junction. 

Experimental Determination of the Thermo-Electromotive Force — 
Temperature Diagram 

The magnitude of the thermo-electric E.M.F. is of the order 
of a few millivolts. It is best investigated by means of a potentio- 
meter of the form described on page 537. An instructive result 
is obtained using the three metals, copper, iron and lead. As 
will be seen from the above, it is only necessary to obtain the 
E.M.F. for two pairs of the metals, the third pair may be estimated, 
making use of the law of intermediate metals. 

A uniform wire potentiometer of, say, 10 metres is employed 
(one of one or two metres, if the other form is not available). If 
the best results are to be obtained a preliminary calibration of 
the wire is advised. 

A steady accumulator, C, is joined in series with R lf R 2 , and 
the potentiometer wire (R x and R 2 should contain resistances 
up to 10,000 ohms). A preliminary experiment gives the value 
of r the resistance of the ten metres of wire. 

Her ice 

Fig. 343 

A thermo-junction is constructed, using copper and iron 
wire. Care must, of course, be taken that the wires are in 
contact at the junction only. For this purpose a suitable form 
of junction is seen in fig. 344. The one wire, B, passes down 


a thin glass tube, G, the other, A, is joined to it at J. To 
ensure good contact, J is dipped into mercury at the bottom of 
the test tube, T. This does not affect the E.M.F. If one or both 
of the metals are affected chemically by mercury, it must be dis- 
pensed with, and special care paid to the welding of the junctions. 
j . Three such junctions are made. One, the hot 
/ / A junction, using iron and copper wires. The free 
ends of the copper and iron wires are each joined 
to connecting leads as illustrated. These two junc- 
tions are maintained at the same temperature, o° C. 
in the experiment described. The use of such junc- 
tions is to eliminate thermal electromotive force 
between the metal of the connecting leads and the 
metals of the thermo-junction proper. In the par- 
ticular case taken, the copper to copper connecting 
lead is not essential generally, but is a safeguard 
against the possibility of impurity in one wire, and 
is necessary for the Pb/Fe junction. 

If the junctions to the connecting wires are main- 
tained at o° C. by surrounding with ice, and the 
' hot junction ' is placed in a water bath at a tempera- 
Fig. 344 ture t o c ^ a;a £jip_ yrfft be developed in the direction 
A to L via the junctions. C is therefore connected with the 
positive pole to B. 

The maximum E.M.F. developed in the above couple is of 
the order of 1500 micro-volts : the rise in potential along AB 
should therefore be arranged not very much in excess of this. 
A voltmeter gives the approximate potential available from 
C, and r being known, the value of R x + R a to cause such a drop 
may be calculated, 

potential difference in AB r 

E.M.F. of C r + Ri + R»' 

the accumulator resistance being negligibly small. 

Having fixed R x and R 2 , the point, L, is made to coincide with 
A, and the standard cell, say a cadmium cell, is put in series with 
the galvanometer ; R 2 is adjusted, keeping R x + R 2 at the 
value determined above, until no deflection is given in G when 
contact is made at A. 

If the room temperature is approximately 20 C. the E.M.F. of 
the cadmium cell is 1-0184. Hence the drop of potential along 
AB is : 

£- x 1-0184, 

whence the drop per cm. of the wire is : 

^- x 1-0184 X I0 " 3 volt • — — — ~ (3) 


The thermo-junction is now placed in circuit and the hot 
junction is raised to about 95 C. in a water bath : the length 
of wire required for balance is obtained. The water is allowed 
to cool and balance points obtained for intervals of temperature 
of about 5 C. 

The hot junction is placed in a hypsometer and again the 
E.M.F. is balanced : when the junction is at the temperature 
of the steam a steady balance point is obtained and noted. 

The hot junction test tube is next placed in a boiling tube 
containing mercury. This is heated slowly. Balance points are 
obtained at intervals and the temperature noted on a special 
mercury thermometer which reads to 360 C. The process is 
carried on until the mercury boils. 

The results are tabulated — temperature and lengths for balance. 
The lengths are converted to E.M.F. by multiplying by the 
reduction factor given in (3) above. 

The process is repeated, using a lead wire in place of the iron, 
and the results are plotted as in fig. 341, E.M.F. in micro-volts 
(io - 6 volt) against the temperature of the hot junction. 

From these results obtain the corresponding curve for lead — iron. 

It will be found that at a temperature of about 240 ° C. the 
Fe-Cu junction will give a maximum E.M.F. This tempera- 
ture is called the neutral point. 

The same results could be obtained directly, using a direct 
reading potentiometer. With a ten-metre instrument r would 
probably be just greater than 10 ohms. If a shunt, S, were 
placed between A and B (fig. 343) and S given a suitable value 
which can be calculated from a knowledge of r, the wire and 
shunt may be made to have exactly 10 ohms resistance. 

As before, suppose the E.M.F. of the cadmium cell were 
1-0184 at tne temperature of the experiment (any other value 
can be treated in the same manner). Then to obtain a drop of 
10- 6 volt per cm. of bridge wire R 2 is given the value (1018*4 — 10) 
1008-4 ohms. The cadmium cell is placed in series with the 
galvanometer, and the sliding contact, L, is made at B (fig. 343). 
Ri is adjusted until a balance is obtained. Then the drop in 

potential from A to B = 5 — of 1-0184 or I0 ~ 8 volt, hence 

the drop per cm. is io -6 volt. 

The lengths for balance in the thermo-couple experiment now 
give the potential in micro-volts directly. 

The same process applies in the case of a single-metre poten- 
tiometer. The value of r is found, and the shunt value, S, to 
make the value for the two in parallel 1 ohm, is calculated and 
R 2 made (1000 E — 1). In that case each millimetre of the 
wire corresponds to a potential drop of io~ 6 volt. 


If no standard cell is available the value of the potential per 
cm. of wire may be calculated from a knowledge of r, R x and 
R 2 and E, the E.M.F. of the cell, C, as determined by means of 
a high-resistance voltmeter. 

Consider the curves obtained. If we take any two fixed 
temperatures and determine from the curve the value of the 
E.M.F. developed, in micro-volts, we shall be provided with two 
equations having two unknown constants, for we have seen that 
the relation between the E.M.F. developed E, and the temperature 
t° C. is : 

E = at + bt*. 

So by choosing two such pairs of temperatures and finding E 
on the Fb-Cu and Fb-Fe curves, the values of a and b for 
iron and copper may be obtained. 

Hence since ' -=- = a-\-2bt, 


we know the value of the thermo-electric power, -=--, at any 



The thermo-electric power lines for Cu and Fe against the 

standard, lead, may thus be drawn and the point of intersection, 

which gives the neutral temperature for iron-copper, may be 

ascertained. This will be found, of course, to agree with the neutral 

temperature as found directly from the E.M.F. temperature curve 

for the iron-copper junction. 

The Use of a Thermo-Eleetrie Couple as a Thermometer 

It will be seen from the curves obtained (fig. 341), that a couple 
such as copper-iron is not a suitable one to use as a thermometer. 
The neutral point is too near o° C. for such use. 

The ideal couple for such purpose is one with a neutral point 
well removed from o 6 C, and which will therefore give an E.M.F. 
approximately proportional to the temperature difference. In 
many experiments in this book such a couple is required. A 
couple of copper and eureka or constantan serves well for 
this purpose. 

When the thermo-junction is used in this way the form of 
potentiometer described on page ^ 539, fig. 339 is a great 
convenience, but the form of direct reading potentiometer of 
ten metres of wire (page 544) when once adjusted serves quite 

In either case the electromotive force-temperature curve of 
the junction should be obtained as described up to, say, 240 C. 
The value of the temperature corresponding to any other electro- 
motive force may then be obtained from the standardizing curve 
so obtained. 



Lippmann's Capillary Electrometer 

The essential feature of the Lippmann Capillary Electrometer 
is a very fine capillary tube drawn from a thin glass tube, mounted 
so that a column of mercury of variable height may be placed 
over the meniscus of a liquid which rises in the capillary. 

Fig. 345 

Fi g- 345 shows how this is usually arranged : C is the capillary 
attached to a vertical glass tube, AB, by means of a short length 
of pressure tube at B. The side tube is connected by pressure 
tubing to an adjustable reservoir, R, full of mercury. A scale 
in millimetres is placed behind either AB or the reservoir R, 
so that changes in level of the mercury may be measured. 

The capillary tube, which dips into sulphuric acid, should be 
of sufficiently small diameter to prevent the acid from being 
driven from the tube by the mercury above it. 

At the bottom of the beaker, E, which contains the acid, is a 
layer of mercury, into which a wire, passing down the centre of 
a glass tube, D, may be placed. 

A second copper wire dips into the mercury above the meniscus, 
either by means of a platinum wire lead fused into the tube AB, 
or by inserting a wire into the mercury in AB or R. The two 
leads are connected to a potentiometer which consists of two 
Post Office boxes in series, R x and R 2 , so that, maintaining R t + 
R 2 at 10,000 ohms, and adjusting R x and R 2 , any fraction of the 
potential of the cell, E, may be applied to the junction of the 
upper mercury and the acid. 



If the meniscus is focussed and made to coincide with the 
horizontal cross-hair in the focal plane of a high-power microscope, 
it will be found that when a potential is applied in one direction, 
the meniscus descends, and when applied in the other direction 
the meniscus ascends. The effective value of the surface tension 
of the acid mercury surface is altered by the applied voltage. 

Suppose the reservoir, R, is lowered so that only a small 
pressure is applied to the surface, and the meniscus is focussed 
on the cross-hair of the observing microscope (which should be 
of high magnifying power), the observed level does not corres- 
pond to the true surface tension level. 

At the junction of two liquids there is a contact difference 
of potential set up, and the value of the effective surface tension 
depends on this contact potential. In whichever direction this 
contact potential acts, the result is a decrease in the effect 
surface tension value. For two given liquids, i.e. sulphuric acid 
and mercury, the value is fixed in direction and magnitude. 

We may regard the surface of separation of the liquids as a 
double layer functioning like a condenser, and having an energy 
per unit area of £cV 2 , where c is the capacity and V the value 
of the contact potential. 

Thus, if T is the effective surface tension, and T the value of 
the surface tension if no potential exists, we may write : 

T=T -£cV2 (4) 

so that whatever sign be given to V the value of T < T . 

If a potential be applied to the surface of separation in the 
same direction as the contact potential, then V increases and T 
becomes less, causing the column of acid to descend to a new 
equilibrium position, whereas, if the applied potential be of the 
opposite sign to the contact potential, we obtain a larger value 
for T, and consequently the acid rises in the tube. 


- Fig. 346 

The capillary when drawn out from the glass tube is always 
slightly conical, and the state of things showing two positions 
is seen in fig. 346. 

If the applied E.M.F. opposing the contact potential increases 


T becomes larger, until, when the applied potential is further 
increased, we have the double layer effect again coming into 
play, due to a net potential difference of the opposite sign. 

For each value of the applied opposing E.M.F. the meniscus 
will take up a definite position, and it is obvious that when the 
contact potential is just neutralized, the meniscus will be at the 
highest point. 

The conical shape of the capillary tube, with resultant change 
of diameter and of focus, renders unreliable the observations of 
the level of the meniscus. To be sure that the diameter of the 
tube at which the surface of separation lies is the same, the 
meniscus is always observed at one point. 

For example, as the value of T increases the level is brought 
back to the previous one by raising the reservoir R and increasing 
the pressure on the surface. The microscope used for observing 
the level of the meniscus should be high-power and the reservoir 
readily adjustable, so that the level of the mercury in R or AB 
may be read for each value of the applied potential. 

This process should be repeated for all values between o and 
2 volts, or until electrolysis interferes with observation. 

The values of the applied E.M.F. should be plotted as abscissae, 
and the pressure in cms. of mercury as ordinates. 

The form of the curve as obtained from the results under, is 
seen in fig. 347. The maximum of the applied pressure^corre- 

Fig. 347 

sponds to a maximum, T, or if the value of T may be represented 
in some such form as (4) the maximum T is T when the term 
depending on the contact potential is zero., i.e. it corresponds 
to an applied potential equal to the contact potential. From a 



knowledge of the direction of the applied potential we may say 
at once which of the liquid is electro-positive and which electro- 
negative to the other. » t . 

In performing the experiment the first thing to do, having 
assembled the rest of the apparatus, is to draw out a suitable 
capillary. After one or two attempts a suitable one will be 
obtained, which is of sufficiently small dimensions to support 
the pressure. 

It should be noted that the capillary tube should be drawn 
from a clean tube. This may be obtained by boiling the tube 
in nitric acid and rinsing in tap water, or by leaving the tubes 
to be drawn in a solution of potassium bichromate and sulphuric 
acid for 12 to 24 hours and then rinsing in tap water. 

The mercury also should be cleaned, preferably redistilled. 

The following is a record of an experiment where the above 
precautions were observed. ? 

Rj + R 2 = 10000 ohms. 






cms. Hg 



























Contact potential from curve = -57 (fig. 347)- The negative 
terminal was connected to the upper mercury, i.e. in the experi- 
ment the mercury was electro-positive to the sulphuric acid. 




(1) Deflection Method 

Suppose Ci and C 2 are the condensers, of capacity K x and K 2 
and that to each is imparted a difference of potential equal to 
E volts. Let £>! and Q 2 be the respective charges on the plates. 

I hen we have : 

K —Sdl If _ Q*. 
Kl - E ' K » ~ "E ' 

or El „ 2l / T , 

In the experiment described below the quantities, Q x and Q 2 , 
are measured by discharging the condensers in turn through a 
moving-coil ballistic galvanometer. In such a case, if B x and 2 
are the angles of the first throw of the ballistic galvanometer, 
corresponding to movement of the spot of light, d x and d g cms., 
from the zero on the scale, we have (see page 479) : 

q, - 1 . 1- ?tf 1 + *Y 

Wl 7T G 2\ 2/ 

Q,_J .i.i»( I+ i\ 

*" 7T G 2 \ 2/ 

or $i = -i, 

If the deflections are small : 

e t _ 2 tan At _ tan 20 t _ rfi 
2 ~ 2 tan 2 ~ tan 20 2 d a ' 

whence Ml = -i ^ _ - 1 , from (1) 

Q 2 *2 K, 

KT^iT- — (2) 



If the deflections are not small this approximation cannot be 
used. The values of X and 2 may easily be obtained, for 

tan 20! == y 1 ' where L is the distance from the scale to the mirror 

of the galvanometer ; hence 20 x and B v 

It often happens that with a steady source of potential, say, 
a steady 2 volt accumulator, that the throw 6 X and a is too 
large, i.e. E is too great. Under such circumstances, instead of 
using E directly we may take any fraction of E by a potentiometer 

Connect E in series with two resistance boxes, R x and R 2 , 
making R t -f R a = 10000 ohms. From the ends of R t connect 
up the leads which in the diagram (fig. 348) are shown directly 
connected with E. 

By adjusting R lf making R x + R 2 = 10000, we may obtain 
a suitable fraction of E, for the potential drop through the 

resistance, R 1( is 


E. These resistances are then 

Ri + R t 
kept constant and the throws are obtained. 

The scheme of connexions is shown in fig. 348. K x and K a 
are condenser keys. K is a single-way tapping key, useful in bring- 
ing the galvanometer to rest after observing the swings. BG 
a ballistic galvanometer of the moving-coil type. 




Fig. 348 

K x is depressed and the condenser, C lt is charged. Kj is now 
raised. The charge, Q lt is thus sent through BG, and X noted. 

As soon as the galvanometer is brought to rest, the key K 2 is 
depressed for the same short time and then raised, a being 
noted. - During the interval of the two depressions of the keys, 
the potential difference between the poles of the accumulator 
will remain sensibly constant. 

Of course the experiment could be done using one condenser 
key, and placing C x and C 2 in turn in the single circuit. 

The observations are repeated, say, four times with Cj and 
three times with C 2 , and the mean values taken. 


(2) Null Method, (de Sauty's Method) 

For this method the two condensers, C t and C 2 , are arranged 
as two arms in a Wheatstone net ; R x and R 2 , two adjustable 
high resistances making the bridge complete. Fig. 349 gives 
the scheme of connexions. G is a high-resistance galvanometer 
and E a battery of cells having an E.M.F. of several volts (say, 
4 or 6 Leclanche cells). 

FIG. 349 

The key, K, may very well be an ordinary condenser key, so 
that when depressed the condensers are charged, and when 
raised the condensers are discharged. The values of R t and 
R 2 are so arranged, that when the key K is moved up and 
down, charging and discharging the condensers, there is no move- 
ment of the galvanometer coil, i.e. in both processes the potential 
at B is the same as at D. 

Let E 1 be the potential at A, and 

v x be the potential at B at a time, t, 
v 2 be the potential at D at a time, t. 

The drop in potential along R x is (E 1 — v x ) and along R 2 is 

{E 1 - v 2 ). Thus, the currents in R x and R 2 are ^-^ and ^"^ 

K t K 2 

In a small interval of time, dt, the quantity of electricity 

passing along R t is ( ~ Vl \ dt, and along R 2 is ( E * ~ ^2 

In charging the condensers the quantity, / III!- 1 • dt, flows to C, 

J Ri 

and f ~ v * -dt to C r 



Now when the condensers are charged, since the potential 

difference is E (that of the cells) we have : 

/jri v 
T5 L * &» 

and similarly 

K 9 E 



E 1 - v, 



where K x and K 2 are the capacities of the two condensers ; 
or K^E = /(Ei - v x )dt, 

K 2 R 2 E = /(E 1 - v 2 )dt, 

when no flow occurs through the galvanometer, i.e. v x = 
throughout the charging, 

i.e. /(E 1 - vjdt = /(Ei - v z )dt ; 

or, KiRiE = K 2 R 2 E, 

ie Ki-^1 

K 2 -R x 

In performing the experiment R x and R 2 may be obtained 
sufficiently high by using a Post Office Box for each, and following 
the scheme of connexions shown. 


Fig. 350 

(3) Method of Mixtures 

Comparison of the capacity of two condensers may be made 
in one or two ways using the 'mixture' method. In this particu- 
lar method the condensers are arranged to have equal charges 
*and the potential required to do this is measured or compared 
by a potentiometer method. The theory of the method presents 
no special points which require separate treatment. 


A battery of cells (two accumulators) is connected as in 
fig- 350 to send a current, c, through two high resistances, Ri 
and R,, which are adjustable and connected together at a point, B. 

The potentials between A and B, B and C, are proportional to 
Ri and R 2 . 

B is connected, as shown, to the two condensers whose other 
plates are connected to the central cups of a cleaned ebonite 
Pohl commutator from which the cross-connexions have been 
removed (or a slab of clean paraffin wax with six holes full of 
mercury will serve). 

A galvanometer, G, is connected to the cups, X, X 1 , of the 

A and C are joined to Z and Z 1 . 

When connexion is made (by the rocker of the Pohl commutator) 
between Y and Z, Y 1 and Z\ the condensers, C t and C 2 , are 
charged to the potential differences of AB and BC, respectively. 

If Qi and Q % are the charges, and K t and K 2 the capacities, 
we have : 

K «=fir K - = fc- -(4) 

c being the current through R x and R 2 . 

When the switch is thrown over so that Y and X, Y 1 and X 1 
are connected, the condensers are discharged through G. 

Suppose the current flow from A to B to C in the potentiometer 
circuit, then the inner plate of C x is at a lower potential than 
the outer, when charged, whereas for C a the inner plate is at 
a higher potential than the outer. So when the condensers 
discharge through G, C 2 will send its charge in opposition to C v 

The above process is repeated, altering Rj and R 2 between 
observations, until finally, a value of R x and R 2 is obtained, 
such that on discharging, no current passes through the galvano- 
meter. Under such circumstances, Q t = Q 2 . 

From equations (4) above we have : 

Q 1 = K 1 R 1 c, Q 2 = K 2 R 2 c. 

In the adjusted position, since Q x = Q 2 , 

KjR^ = K^, 

K, R 2 
or — - = — =• 

K 2 Rj 

Determination of the Absolute Capacity of a Condenser 

The value of the capacity of a condenser may be determined 
in any system of units if a measured potential, E, in these units, 


applied to the condenser imparts Q units of charge in the same 
units ; for the capacity, K, of the condenser is denned by Q =K-E, 

or K=2 (5) 

For example, if Q is expressed in coulombs and E in volts, K» 
the capacity, obtained in the equation above is expressed in 
farads. This can be transformed to micro-farads since this unit 
is 10- 6 of the farad. 

The usual method of finding K experimentally is to apply a 
known potential (in volts) to the condenser and measure the 
charge, Q, by discharging the condenser through a ballistic 

The moving-coil ballistic galvanometer is the best form to 
use, as in most capacity experiments, whence if a! be the first 
observed throw of the instrument, due to the discharge, we 
have (see page 479) : 

or, if the method of page 480 is used, the value of x is not obtained 
so fully. If ax and a 8 are the first and second displacements of 
the spot of light on the same side of the zero, we have : 

it 9 2 \a 3 / 
when <p is the angular deflection produced by a steady current c. 





' ^ ■ 






Fig. 351 

For small displacements of <5, d lt and d 3 cms. on the scale corres- 
ponding to <p, a, lf and a 3 , we have : 

O = - .-.^±( dl 





Fig- 35 1 shows the scheme of connexions for such an experiment. 
K is the condenser whose capacity is to be determined ; R x and 
R 2 are resistance boxes, introducing a high resistance to the 
circuit ; AB a resistance box of decimals of an ohm, r say ; 
K 1 a single-way switch ; K 3 a tapping key, whereby the galvano- 
meter may be brought to rest ; K 2 a condenser key ; and BG a 
ballistic galvanometer of the moving-coil type. c- 

The circuit, R 1} R 2 , and the battery, E, serve in the first part 
of the experiment as a potentiometer.* R x and R 2 have a 
constant sum of about, say, ioooo ohms. By adjusting these 
two values, keeping (R x + R 2 ) constant the potential difference 
between B and C may be made any desired fraction of the 
potential, E, of the battery. This adjustment is carried out, 
and the deflection produced in the galvanometer when jthe 
condenser is afterwards discharged through it is noted. 

Since R t + R a is high and r is never greater than I ohm, the 

potential between BC = = — -~-*'E=y volts, say. 

Ri + K 2 

When K 2 is depressed, the condenser is charged, the potential 
applied being V. The charge, Q, will flow through the galvano- 
meter causing a deflection, d x cms., when K 2 is raised. 

This process is repeated, say three times, and a mean value 
of & x obtained for the fixed potential, V. 

For each discharge, the reading corresponding to the second 
deflection, d 3 , on the same side as d x , is measured. This 
provides the data for the damping correction for the 
galvanometer under the identical conditions under which oc 2 
is measured. 

The time of swing, T, is obtained by timing, say, 20 swings 
of the needle with a stop-watch. It is most convenient to use 
the position of rest of the spot of light as a reference point of 
such counting and timing of swings. 

To standardize the galvanometer a steady current of known 
magnitude must now be sent through it and the scale deflection, 
8, measured. 

The key, K lt is now closed. The current from the battery E 
(the same 2 volt accumulator throughout), now passes through 
R 2 and R x , and then through the galvanometer, shunted by the 
small resistance, r. 

The resistance of the galvanometer and shunt is : 


G +r 

where G is the resistance of the galvanometer, t 

t G may be obtained by including a resistance between B and BG, and adjust- 
ing till the deflection is reduced to half the value, when G is obviously equal to 
this resistance. 


Hence, if E is the potential difference in the circuit, the current 
in the main circuit, *, is : 


R 1+ R,+B+Jg; 

where B is the resistance of the battery. This may be neglected 
when the latter is an accumulator, and — ^is > 1 ohm, and is 

also negligible compared with R + R 2 - 

Of this current, the value of the part, c, through the galvano- 
meter is : 



(r + G)(R 1 +R 2 ) 
We have also seen that the potential applied to the capacity is 

V = 5i — • E. 

R x + R 2 

Hence, substituting this value of V and Q from (6), 



2* (r + G) (R, + R 2 ) 

d JL(<h\* 


Ri + R 2 

^ILftA* ( r V 

27T d \dj XRfi) 
if r is negligible, cf. G. E is assumed constant throughout the 


Fig. 352 

Measurement of Capacity, using a Fluxmeter 

A simple method of estimating the capacity of a condenser 
is illustrated in fig. 352. 


The condenser, C, of capacity, K x , is connected to two condenser 
keys, Si and S 2 . The lower studs of the keys are connected to 
the mains of the electricity supply (i.e. to 100-200 volts). When 
the keys have been depressed the condenser is charged to that 
potential, V. On releasing the keys the charge on the condenser 
is discharged through a low resistance, r, which is shunted across 
the fluxmeter, F (the position shown in the diagram). 

r is made small, about ^ to ^ ohm is suitable, and the shunt 
may therefore be regarded as taking the whole of the discharge, 
say, Q units. 

The potential difference at the ends of r varies from V, to zero. 

i.e. the current through r at any instant when the potential is E 

is — and the total quantity of electricity passing through r is 

therefore : Q = f— ■ 

Now, if the resistance of the fluxmeter, usually about 30 ohms, 
is large compared with r, the instrument gives a deflection which 

is a measure of /E • dt (see page 484). Let this deflection be 

x divisions or x x 10 4 maxwells, 

o — x x iq4 

W — f 

If now r is expressed in ohms, the units must be made consistent, 

x x 10 * 
i.e. a; xio* maxwells should now be written ^ — practical 

units to obtain Q in coulombs, 

i.e. Q (in coulombs) = 

If the potential applied to the condenser is measured by 
means of a voltmeter, the capacity in farads may be readily 

The Effect of Inductance and Capacity in a Circuit Conveying an 
Alternating Current 

In many of the subsequent methods of finding the self -induct- 
ance, mutual inductance, or capacity in a circuit, alternating 
current is employed and a system of non-inductive resistances, 
capacities, inductances are arranged to produce a balanced 
Wheatstone net, the usual galvanometer being replaced by a 
telephone as detector. In such circumstances it is not a difficult 
matter to find the relation between the values of the resistance, 
etc., when such a balance is obtained. The majority of cases 
may be solved by precisely the same method. 


Consider an alternating potential E cos pt to be applied to a 
circuit, of resistance R, self-inductance L, and containing a 
capacity K (fig) 353), (p is, of course, given by 

27m = pt 
where n is the frequency of the alternation). 

'Ofl'OWOO N — — 




E co* yt 

Fig. 353 

Now E^ = E (cos pt -h i sin pt), 

where i = V^i a* 10 ^ tne rea ^ ^ axt °* E^*' expresses, therefore, 
the alternating potential applied. 

If c is the resulting current at any instant the net applied 

potential = Re, 

Lie Q 
i.e. Re =E cos^--^--^, 

when Q is the charge on the condenser at any time, t. Further, 
we have : 

c ~ dt' 




** dt* + dt + K y 

Hence : L ^ + R 3? + ^« = real of E^*', 
If Q = A-e*< , i.e. § = #Ae*< , ^ - - p*A& , 

we have, where A is a constant, 

- L£ 2 Ae*' + R# Ae*< + g Ae* = E e*'; 



A = 

E ( 




R#-I4> a +^ 




c = 

dQ ipE^ 1 

dt Rip -Lp* + ± 

c = 

R + i( L p-^) 
Ea e»<\R-i(Lp-±)} 


R8 + ( L ^- ^) 

. E e» f 



i R2 +(^-^) a r^i Ra +(^-i) 2 } 


1 7 FTm* = cos * 

: R ' + ( L *-^)I 

R 2 + 


Fjl = sm a, 

we have 

c = - 

E n e*' 

The real part of this is] 

^ E cos (pt — a) 

"> ,+ ( i *-i5) , r 


i.e. the current lags behind the applied E.M.F. unless a = o, 
i.e. cos a = I, 


i.e. unless R = \ R 2 + 


or Lp = =jr-, or p =- . 

F Kp' r VK-L 

Thus we see that in the general case taken the method gives 
the value of the current at any time. 
From equation (7) we see that the maximum current C is equal 

to the applied electromotive force divided by ( R + Lip — — Y By 

analogy with the simple Ohm's Law equation ( R + Lip — J 

may be termed the ' quasi-resistance ' of the circuit, and be used 
in the same way as ordinary non-inductance resistance values, 
when alternating electromotive force is applied to a circuit. 

For example, consider a Wheatstone net as illustrated in 
fig. 360. If a direct current were applied at AC and the net 
balanced we have : 

Resistance AB _ Resistance AD 
Resistance BC ~~ Resistance DC 

Now for alternating current applied using the quasi-resistances 
in the corresponding arms : 


r s + Lip R _ * 


which gives on rearranging and equating real and imaginary 
quantities, thus : 

(real) r x R = M or L = K^R, 


(imaginary) ]0 = 1T^ or K i r i = K * r » 

a result which agrees with the fuller treatment given on page 575. 

The above method of investigation assumes that a true sine 
E.M.F. is applied. Using a ' hummer ' or the secondary of a 
small induction coil does not produce such a simple form of 
E.M.F., but would be the resultant of a series of such simple 
wave forms. 

Such an E.M.F. would be satisfactory in those cases where 
the end result, as in the case taken above, is independent of the 
frequency or p. 


In cases where p is involved in the result, a timed telephone, 
which responds to one frequency only should be used. As an 
alternative, a vibration galvanometer such as that designed by 
A. Campbell (" Physical Society of London Proceedings," 1907, 
page 626). As such methods are not very much used, they are 
not included in this account. 


The measurement of the coefficient of self-induction of a coil 
may be carried out in several ways, of which the following 
methods are representative. 

Some of these methods are very tedious as they involve a 
double balance of a bridge, for steady and variable currents. 
Perhaps the most unsatisfactory methods from this point of 
view are (1) the direct comparison with a standard self -inductance, 
(2) Maxwell's method. Rimington's method, too, is apt to 
be tedious. The methods which have been found to be most 
satisfactory in the laboratory are Rayleigh's, Anderson's, and 
Owen's. These methods are therefore used generally when a 
choice of methods is available. 

Fig. 354 

(1) Rayleigh's Method 

In this method the inductance, L, is placed in series with a 
small variable resistance, r, and arranged in the arm, AB, of 
the Wheatstone net, as shown in fig. 354. The remaining arms 
are of the same order of magnitude as the resistance, AB. G is a 
ballistic galvanometer preferably of the same order of resistance ; 
K x and K a are keys. The resistance, r, is reduced to zero and' 
the network is balanced for steady currents by closing the key, 
K 2 , before making the galvanometer circuit through K x . 


To obtain an accurate balance, it may be necessary, when the 
arms are equal, to introduce a smaller resistance than is available 
in the Post Office Box which provides the three resistances, r 8 , 
r z and r A . This may be effected by having a length of platinoid 
wire in series with L, and adjusting the length until an accurate 
balance is obtained. 

If now a small E.M.F., E, is introduced in one arm, a current 
which depends on E will pass through each of the other resistances 
in the network. 

Thus, if the battery key is closed, K a being meanwhile closed, 

an E.M.F. of magnitude L-y is established in the arm, AB, 

resulting in a current in each of the arms in the net. Let the 

current in the galvanometer branch, BD, be k • (Lt), where k is- 

a constant which depends on the value of the resistances. 
Under such circumstances the total quantity of electricity 
which passes through the galvanometer due to this cause is : 

/* dc 
kL -j t dt = kLc , „ (9) 

where c is the maximum steady current flowing through AB. 

This quantity, Q, may be calculated from the observed throw 
in the ballistic galvanometer, and thus L is obtained in terms 
of the constants of the galvanometer, k and c . 

Assuming that a moving-needle galvanometer is employed 
we have, page 477, 

Q -v i • «i (*+!)- «*■ ™ (I0) 

where x is the logarithmic decrement. 

To eliminate k and c , a measurable small potential charge 
is introduced into the arm AB. This is brought about by 
adding a small resistance, r, to AB. (r should be not greater 
than g£o r 1 , usually j^ ohm does very well.) 

Assuming that the current, c , will not be materially affected 
by this small charge, the potential introduced in the arm AB 
is c& : this causes a current, kc^/m. the galvanometer, producing 
a steady deflection, r 

Then G being, as above, the galvanometer constant, i.e. the 

field strength at the centre of the coil due to unit current, we 

have : 

kc rG = H tan Q lt 

, H tan 0« . 

or kc = ~x • * • 

G r 

Substituting this value in equation (10) above, we have : 

0/ X^ 

rp sin -j 

L = r 

T sin M I+ ^ 

n tan0j 

If r is expressed in ohms and T in seconds, L is in henrys 
(io 9 C.G.S. E.M. units). 

If a moving coil type of ballistic galvanometer is used and 
r ? and r 4 or r x and r 3 are small, the galvanometer may give but 
a small deflection as it is shunted by these resistances. This 
is the case when the resistance of L is small. Measurable and 
reliable results can be obtained if the galvanometer circuit is 
broken the moment the discharge has passed through it. This 
condition is most conveniently brought about by using a single 
key as shown in fig. 298 on page 481, for both battery and 
galvanometer circuits. 

The three brass strips, A, B, and C, are insulated and connected 
to separate terminals. When C is depressed contact is made 
between C and B, then A and S, but not between B and A which 
are separated by an ebonite stop. S and A replace K 2 ; C and B 
replace K 2 , when C is pressed down and a steady current flows 
in the circuit, the galvanometer which is in the circuit shows 
no deflection for a balanced bridge. On releasing C, the bottom 
contact is first broken (i.e. the battery circuit), and a very short 
interval of time afterwards C and B are separated. This interval 
is sufficient to allow the impulse to be given to the galvanometer, 
coil which, due to the separation of B and C, swings without 
excessive damping due to induced currents. 

If the moving-coil instrument is used in this way, the end 
result is slightly modified, for in this case (see page 479) 

where is the corrected value of the first throw for no damping, 

i.e. is the observed throw X f J. 

Now for a steady current equal to kc^ the couple on the coil 
is GkCff, so that 

Gkc r = T0J, 

t being, as above, the restoring couple in the suspension, per 
unit angular displacement. Combining the last two equations 
we see that 

n 2 * 0i " 


(2) Comparison with a Standard Inductance 

To effect a direct comparison of an unknown inductance, L^ 
with a standard inductance, L 2 , the arrangement of fig. 355 
may be employed. / 

In series with L x is a non-inductive resistance of a variable 

Fig- 355 

magnitude, say a length of manganin wire, which is adjusted 
so that the total resistance included between A and B is r x ohms. 
By means of a key, S^ the galvanometer, G, is included in the 
circuit to which a direct current from a 2 volt cell is supplied via 
key, S 2 , and a balance is obtained for steady currents, i.e. the 
balance is obtained by closing the battery circuit first, and 
when the current is established, the equality of potential at 
B and D is tested by closing the galvanometer circuit. 

Having balanced the bridge for steady currents the balance 
is now tested for variable current. The first steady balance will 
not be one which also is in adjustment for variable currents. 
The resistance in series with L x is given another value and the 
steady current balance obtained once more for the net, which 
is once more tested with variable current. This process, which 
is often a lengthy one, is repeated until the resistances, r v r 2 , r t , / 4 
are of such magnitude that the points, B and D, are at the same 
potential both with steady and varying currents. 

It will be shown that under such circumstances 

Li = ri = U,, 

The variable current may be obtained by breaking and making 
the battery circuit; or by switches, S x and S 8 , a telephone ear-piece 


may be made to replace the galvanometer, and an alternating 
current from a buzzer supplied instead of the direct current used 
in the steady balance experiment. 

Assuming an alternating E.M.F. be applied to AC and the 
network adjusted so that the potential at B is the same as that 
at D, at all times we have, using in the Wheatstone relation 
the ' quasi-resistances ' of the arms : 

Ljp + r x = Tj 
L/p + r z r 4 
Equating the real parts we have : 

»V« = r 2 r 9 , or £ =~ 3 , (n) 

a condition which was fulfilled by first balancing for steady 

Equating imaginary parts of the equation, we find 

T y *» 

or _-L=— 9 — _J i, from (n). 

(3) Maxwell's Method 

The arrangement of resistances and the inductance to be 
measured is shown in fig. 356. The method involves a comparison 
of L, with a capacity of known value. In the present method 
the resistances r lt r t , and r 8 , together with L, are made into a 
balanced Wheatstone net, using a cell and galvanometer in the 
usual way. In parallel with r x is placed a condenser whose 
capacity, K, is known. During the process of balancing the 
resistances for steady currents, care is, of course, taken to ensure 
that the current is established before the galvanometer key is 
depressed to test the balance. The experiment may now be 
completed either with the aid of the galvanometer across BD 
and the cell connected through a switch to AC, or by means 
of a telephone ear-piece as detector and an alternating E.M.F. 
applied from a hummer.* 

If the galvanometer is retained, the current is made and broken 
whilst the galvanometer is permanently in the circuit, and 
r % and r 8 are adjusted until no deflection is obtained in the 

If the alternating E.M.F. from a hummer is used, the resistances 
are adjusted until no noise is heard in the telephone ear-piece. 

In either case the steady current balance will, in general, be 
upset. This must be again obtained and then the test repeated, 
until finally the balance is equally good for steady and varying 

* It should be noted that the galvanometer and steady E.M.F. may be dis- 
pensed with in experiments (2) (3) and (5). Alternating current may be used 
alone. The first adjustment is to reduce the noise in the phones to a minimum, 


When the network is balanced for steady and alternating 
currents the relation between the four ' quasi-resistances ' may 
be applied. 



Fig. 356 

Since the arm, AB, consists of — -r=~ and r x in parallel, the 
effective ' quasi-resistance ' is 

1 -f- ipKr x ' 
whence : 


r,(i + ipKrJ r 4 + Lip ' 

where r 4 is the total ohmic resistance in the arm, DC. 
Equating real quantities in the above equation : 

r*r« == y,r^ or -^ =— *, 



a condition which was experimentally established in the steady 
current balance. 

Equating imaginary terms : 

r % r z pKr x = r t Lp, 
or L = r 2 r 8 K, 

or from (12) above = r jftK, 

from which L may be calculated. 

(4) Rimington's Method 

A method due to Rimington enables a determination of L to 
be made in terms of a known capacity, making use of a modifica-" 
tion of the Maxwell arrangement. The self-inductance is placed 

and the second to reduce this minimum to zero. N.B. — The value of the resistance 
for intermittent current is only the same as that for alternating current when the 
frequency of the latter is small. 


in the arm, DC (fig 357). The capacity, K, is placed in parallel 
with a variable portion, r, of the resistance r x (AB). The network 
is balanced for steady currents, and then the point, E, is chosen 
such that when the battery circuit is made and broken no 
deflection is produced in the galvanometer. This method has 
the advantage that in the second balance the adjustment of the 
resistances, r x , r s , r s and r 4 , is not disturbed. 

When the above conditions are fulfilled we have the following 
relation between the physical quantities involved. 

L = K — - henrys. 

Fig- 357 shows the arrangement of apparatus. For this 
experiment it is not possible to use a telephone and alternating 
current, as will be seen by working out the case by the general 
method given ; the conditions required for balance with such 
currents reduces the method to Maxwell's, i.e. r x — r = o. 

To establish the formula given above for the case of a balance 

with intermittent direct current, let c x be the steady current 

in AB (resistance, r x ) when the key is closed ; c 4 the steady current 

in DC. Then when the circuit is broken, the quantity of elec- 

Lc . /*° hdc 
tricity which passes through the inductance is -~- i.e./ ^-, 

where R is the resistance between the points D and C (r 4 ) together 
with the resistance of the rest of the net, i.e. 

R = * 4 + r 2 + 

G(r 3 + r x ) 

where G is the resistance of the galvanometer. 


The fraction of this quantity which passes through the galvano- 
y i +r 3 of LC4 
r x + r 3 + G R 

i.e. is 

which reduces to 

LC4K + r z ) 

G(r 3 + r t ) 

G + r 3 + r x , 
Lc 4 r x 

('i + r z + G) 


'i('i + '3) + Gfo + r 2 ) 
Also when the circuit is broken, the condenser discharges a 
quantity of electricity through the galvanometer, in the opposite 
direction to that due to the inductance. 

Now the quantity of electricity on the condenser due to the 
steady current, c v in AB is Kfrcj). 

The part of this quantity passing through the resistances 
other than r is : 

Krc 1 . _ . - 

ri + f » + LG+r, + r; 
Of this amount, that passing through the galvanometer is : 
Kr 2 ci x Jr 2 + rj 

r + r -I G ( y 2 + y 4) (G + r. + ftf- 
1 8 G+r 2 +y 4 
This reduces to 

Kr 2 Cjr 2 


' 2 ( r t + r 3 ) +G(r! + r 2 ) ' 
when substitutions are made using the fact that r#i — r 2 r z . 

When the quantities represented by (13) and (14) are equal 
and opposite, the galvanometer will be unaffected by making 
or breaking the circuit by means of the key shown in the figure, 
i.e. K> 2 CjT 2 = Lc^j, 


We may further express — in terms of the resistance for when 

the currents have acquired the steady values, c x and c 4 ,we have : 

£1 = *z + U . 
c 4 ?i+r 2 ' 

'1 r 2 r x + r 2 c A r 2 

Hence L = ^-^ 


Experimental Details 

The chief disadvantage to this experiment is that much time 
may very easily be spent in a fruitless effort to determine the 
value of an inductance, unless the order of suitable resistances, 
etc., to balance the network, is known. If, however, the approxi- 
mate magnitude of L is known we may, as in the following 
example, obtain this information. 


An inductance (of 98*15 ohms resistance) is to be evaluated, 
using a capacity, one micro-farad. 
Now from above we have : 

L = K -^, 

and the inductance is known to be of the order of one henry, 

i.e. y 4 is of the order 10 2 ohms., 

K is of the order 10- 6 farad ; 

10- 6 
r 1 = — 


x io 2 r 2 = io _4 r a « 


r is a fraction of r v or r = ar lf where a is not greater than 1, 

Zj__- -rn-4 


a 2 r, 

The arm, AB, of the figure could be therefore two Post Office 
boxes in series. To obtain a subdivision as at E, however, it 
will be convenient in this case to use three Post Office boxes, or 
boxes having a similar range of resistances, for this arm of the 
net, as in fig. 358. The condenser, K, is placed in parallel with 
boxes 1 and 2 as shown. 


The balance for steady currents having been obtained, the 
value of r is varied by taking out plugs from 1 and 2, and inserting 
the same value in 3, thus keeping r x constant and of the value 
required fpr a steady current balance, until r is finally obtained, 
such that the potential drop along AE is such as to cause a 
current through the galvanometer equal and opposite to that 
due to the inductance. Values of r too small will be shown by a 
movement of the galvanometer to the one side due to the induct- 
ance effect predominating, whereas when c is too large the 
capacity effect will be large and cause a deflection in the opposite 

This process was employed to find the value of L for the coil 
mentioned above, and the following values obtained, 

r x = 20000 r z = 1963, r = 14500 

r % = 1000, r t = 98*15, K = 1 micro-farad, 

K-lxiHS'xio-xgiis _ x . 03 henrys- 

whence L = 


The self -inductance was a nominal 1 henry. 

(5) Anderson's Method 

The scheme of connexions shown in fig. 359 illustrates a very 
convenient method of finding the value of an inductance in 
terms of capacity and resistance. 

Fig- 359 

The Wheatstone net, r t , r 2 , r 3 , r 4 (r 4 being the total resistance 
of the arm, DC, including the self-inductance) is arranged as 
shown. The condenser, K, is connected from A to a variable 
point, E, in a resistance r (BE). 


The network is first balanced for steady currents using a 
battery between AC and a galvanometer, G, between ED. For 
such steady currents K and r do not have any effect. When a 
balance is obtained, by first depressing the battery key and then 
the galvanometer key, we have : 

^i = - 3 . 

The network must now be balanced for varying currents. 
This may be done in one or two ways : either by using the 
galvanometer and an intermittent direct current, or an alternating 
current and telephone receiver. If the galvanometer is used 
as a detector, a tapping key is inserted in the battery circuit 
and r is so adjusted for the particular value of K, that on making 
and breaking the circuit no deflection of the galvanometer is 

When this condition holds, it may be shown, by the application 
of Kirchoff's Laws to the circuits, ABED A, ABCDA, that 
L = K{fy, + r(r,+r«).} 

The more convenient way of proceeding with the experiment, 
however, is to replace the galvanometer by a telephone and 
the battery by a source of alternating current. In the general case, 
when these changes have been carried out to conform with fig. 
359 a note will be heard in the ear-piece when AE is closed. 

By adjusting r (the resistance BE) this note may be reduced 
in intensity to zero. 

We may readily find the relation which exists between the 
various resistances, L and K, under such circumstances. The 
condition satisfied is that the potential drop between A and E 
is the same at all times as the potential drop between A and D. 

Let the currents in the arm, AB be c- l ; in AE, c 2 ; and in 

AC, c. Then, when no current passes in either direction along 

ED, the magnitude of the current along BC is c x + c 2 . Consider 

the circuit ADC, and let the applied potential be taken as the 

real part of E e*', The potential difference between A and 

D is r z c, where c is the solution of 

E e*' = r z c + Lj t +r 9 c. 

Putting c = he*** in the above equation we have : 


U) + 
E e^ 

A = E o ; 

whence c = -. ; — » , T . . » 

and potential difference between A and D is therefore 

E e** 
r a +r] + Lip' r * ••' ( I5 > 


Now let Q be the charge on the condenser, K, at any instant. 
From a consideration of the circuit, ABC, we have : 

E^*' = r x c x + r 2 (c x + c 2 ) (16) 

But the potential drop along AB is the same as the drop 
along AEB, or 

Q / * 

ClTl = K + c *' " * ' ^ 

Substituting for c x (17) in (16) and putting c 2 = -J^, we have : 
whence as before : 

Q *£L 

from which the potential difference between A and E ( ^ J is 

p Eof^ (18) 

■( I+ &)+K*(, i+ a + r ) 

The condition specified above for no current through the 
telephone is that (15) and (18) are the same for all values of t. 
Equating these expressions we have : 

('. + U + Lip) E«** = f ,|(i + £j) + ^(fif .+Wi+frJ JE^'; 

whence equating the real parts we have : 

r. + ,.-,.(*±r,). 

or — = — , 

a condition which is assumed by the initial balance for steady 
Equating the imaginary quantities we have : 

Lp = ^Kfor, + rr x + rr 2 ), 

or L = K{r z r s + r(r s + u)} . (19) 

This result could have been obtained using the quasi-resistance 
method already outlined. For we may consider the circuit taken 


above and write down the complete ' Ohm's Law Equation ' and 
find c,c x and c 2 and hence the potential drops, AE and AD, and equate. 
Thus, for circuit ADC : 


'* + y 4 + L # 

ial drop, AD = 
Circuit ABC 


Potential drop, AD = r * B ° e *\ (20) 

r z + r * + ^P 

E e** = e 1 r 1 + (c s + cjr s (21) 

By Kirchoff' s Law since the E.M.F. in circuit, AEBA, is zero, 

Substituting in (21) above, 


and since ^f=*c Xi we have, integrating, Q = ^ and the potential between 

AE, which is g., is 




Equate to the value given in equation (20), we have, equating the 
real quantities : 

'* u 

and equating the imaginaries, 

I* = K{r s r t + r(r t + r 4 )}. 

(6) Owen's Method 

The following method of finding the value of the self -inductance 
of a coil was described by Dr. D. Owen in the " Proceedings of 
the London Physical Society," 1914-15, vol. xxvii. 

The coil whose inductance, L, is to be found is placed in one 
arm of a Wheatstone network, in series with a variable non- 
inductive resistance, so that the total resistance is r 2 ohms. 
The scheme of connexions is shown in the diagram (fig. 360). 

K x is a standard capacity, K a a second capacity, and R an 
adjustable non-inductive resistance. 


It will be noticed that in this method there is only alternating 
current supplied to the bridge. In the first case the resistance 
R is made equal to zero, and the resistance r 2 is adjusted so that 
a minimum intensity is heard in the telephone. R is then 
gradually increased until this minimum is reduced to silence. 
Under these circumstances the relation 

L = K^R^K^R 

is found to hold. 

Fig. 360 

Now it was shown on page 561, as an example of the ' quasi- 
resistance ' method of solving such problems, that when such a 
balance is obtained 

L = K^R = K> 2 R. 

As a contrast with the method given on that page, we may 
show that /the above expression gives the relation between the 
capacity and resistances used, from a fuller consideration. For 
let the applied E.M.F. be E^*'; and, since no current flows 
through the galvanometer in the balanced state, let c x be the 
current in ABC. 


For the circuit ABC we have : 

E^ = Cl (r 1+ r 2 )+L^; 

hence, putting o x = Ae*' and substituting to find A, we have : 


'* ^ + r 2 + Lip' 
The potential difference between A and B is r x c x or 


r% + U + L# 

Now, if c 2 is the current in ADC, we have for that circuit : 

E^ = Rc 2 + Q(^+^-), 



or, since -jr= c lt 

E ^= R f+Gt + r> 

gives the following 

Hence putting Q = Be* 1 and -^ = B#d?*' we find B, which 

Q ** 

The potential difference, at any time, t, between A and D, is 
therefore ~- 

_x_ E g»* 

The condition that there is no sound in the telephone is that 
at all times, t, the potential at B = potential at D, i.e. for all 
values of t : 

i.e. r&^Rip + (± + i) | = (r x + r t ) + Up. 

Equating real quantities : 

' iK ' it + r.) -'• + '• 


Ki + K 2 _ r^jfyg 
K 2 r x * 

K 2 tr 
or rj&i = rz& 2 

Equating imaginary quantities : 

or L = /iKjR or r 2 K 2 R 


Direct Measurement with a Ballistic Galvanometer 

The coefficient of mutual induction of two coils may be denned 
as the number of lines of magnetic force which pass through one 
coil when unit current circulates through the other. 

Thus, if a current of maximum strength, c , passes through 
one of the coils, whose mutual induction is M, Mc lines of mag- 
netic force thread the second ; and whilst the current grows 
in the primary, the number of lines of magnetic force threading 
the secondary is changing. Therefore an induced E.M.F. is set 
up in the secondary during the time of growth of the primary 
current. This E.M.F. is numerically equal to the rate of change 
of the number of lines of magnetic force in the secondary, i.e. 

= — (Mc)* where c is the instantaneous current in the primary 

during the growth of that current. 

If L is the coefficient of self-inductance of the secondary coil, 
and c 1 is the current in the secondary corresponding to c in the 
primary, we have a further E.M.F. in the secondary due to the 

hdc 1 
self-inductance numerically equal to -tj—> i- e - if R is the total 

resistance of the secondary coil circuit, neglecting signs, 

Rcl = L Tt + M Tf 

Now Q, the quantity of electricity passing through the second- 
ary, is / cHt where the integration is carried out over the whole 
time during which c rises to the steady value c . 

Q =jcHi -=f~dc^ +f~dc. 

* N.B. — This only applies to coils with non-magnetic cores. For if there is 
an iron core the value of the flux is not proportional to the current. 




= and 

The value of c 1 at the commencement and the end of this 
integration is zero, hence : 


If, therefore, the second coil is connected to a ballistic galvano- 
meter and the throw is 0, due to the passage of this quantity of 
electricity, M may be calculated in terms of R, c , the constants 
of the galvanometer, and 0. 

In this case it will be well to calibrate the galvanometer in 
the circuit since R includes the resistance of the galvanometer 
and the second coil. 



Fig 361. 

Fig. 361 shows a convenient disposition of apparatus to carry 
out a direct measurement of M on these lines. 

The current in the primary coil may be regulated to a suitable 
value by means of the resistance R x . In series with R x is a small 
resistance r(A»). C is a four-segment commutator, or may be a 
switch consisting of four mercury filled holes in a block of paraffin 

K 1 and 2 are connected together the ballistic galvanometer 
is in direct circuit with the secondary coil, and when K is depressed 
a deflection, 0, is obtained for the establishing of the steady 
current, c , in the primary. If a moving-needle ballistic galva- 
nometer is used, we have 


H . 0„ Mc 
G 2 



where d is the value of the first deflection corrected for damping, 
i.e. O = f i + - 1 where x is the logarithmic decrement and 
is the observed throw.* 

* If a moving-coil instrument is used, the form of double key, described on 
page 481, should be used ; K is replaced by the lower pair of contacts in such a 
key and the upper pair act as a key in the galvanometer circuit. 


If now C is arranged so that connexion is made between 
1 and 3, 2 and 4 only, and a steady current, c , is passed in the 
primary circuit, the potential drop established at the ends of r 

is C&, i.e. the current through the galvanometer is -^ » smce r is 

very small compared with the resistance of the galvanometer. 
If this causes a steady deflection, lt we have : 

G. c ^=Htane 1 , (26) 

as shown on page 481. 

Combining (25) and (26), we obtain the following value for M : 

it T Smi 

If a moving-coil galvanometer must be used, since Q =-• ^ • — 
the value for M becomes : 

M = 

J . 1 . 1°. 

w 2 2 

It will be noted that the value of R is eliminated : r may well 
be a standard ^ or jf ohm. T is obtained in the usual 
way by timing 20 or 30 swings, under the conditions of damping 

which o.btain during the observations above ; O = 6 1 1 + - j 

The value of x may be obtained by one of the methods given 
on page 480. 

Comparisons of Mutual and Self-Inductance. (Maxwell's Method) 

The one coil of self-inductance, L, is arranged in one arm of 
a Wheatstone network ; the other coil of the mutual inductance 
is connected, as shown in fig. 362, in series with the battery. 
For this method the resistance, r (AC), is not present. The coils 
are arranged by trial so that the self and mutual induction effects 
in the arm AB are opposed. 

The resistances, r lt r t , r z , r 4 , are balanced for steady currents. 
The battery circuit is then made and broken, the galvanometer 
circuit being closed. r z , r % , r 4 are adjusted until the minimum 
effect is produced in the galvanometer for such current changes. 
The steady balance is again tested. By repetition the values 
r v r z , r 4 , may be finally arranged such that for steady and varying 
currents the galvanometer remains unaffected. 


The steady current balance gives in the usual way the relations 
between the resistances, 


During the fall of the current at break, the self-inductance 

effect is an E.M.F. L^- 1 , where c t is the current in the arm ABC. 



In the opposite direction an E.M.F., M ^-, is set up in the same 

arm due to the mutual inductance. If these two E.M.F.'s are 
equal and opposite, 

L-^ = M jt (numerically), 

where c = c x + Cj, 

Lc x = M(c x + c 2 ), 
L c x +c a 
M c x ' 

-* + '? 


It is apparent from the above equation that the experiment 
may only be performed if L is larger than M. 


This method is open to the same criticism as the Maxwell and 
comparison methods of finding L, and is therefore seldom used. 

A modification of this method is obtained by the insertion 
of the arm, AC, of a variable resistance, r, as shown in fig. 362 

With K 2 open a steady current balance is first obtained. 
This having been accomplished, the resistances, r v r 2) r z , r 4 , are 
fixed ; K 2 is now closed and r adjusted until the galvanometer 
is unaffected by intermittent current in the network, e.g. when 
the current is made and broken or reversed. 

As before, no resultant E.M.F. is set up in AB during such 
changes of current, otherwise a current proportional to such 
E.M.F. would pass through the galvanometer. Thus we have 
again : 

I^i =M^ (numerically), 
at ctt 

where c is the current passing through the coil M, i.e. is 
{c x + c 2 + c 3 ). 
Integrating as before, we have : 

LCi = Mc, 

L _£_ 

1,e# M c/ 

Now if R is the resistance of the whole net between A and C, 
we have : 

cR = c 1 {r 1 + r 2 ) 

L r x + r 9 
M~ R ' 

= r(r t + r t ) + r{r 3 + u) + {r x + r t ) (r 3 + r 4 ) 
r{r 9 + rj 




+ (rx + rj 

M r 

L = r{r 1 + r 9 ) + (^i + r 2 ) r s 
M rr z 

This arrangement of resistance eliminates the trouble- 
some double balancing of the last method. The steady balance 
is obtained with a cell and galvanometer in the usual way. 
The battery circuit is then made and broken or reversed ; r is 


adjusted until the galvanometer shows no deflection during that 
process. As before, care must be taken that the coils are so 
connected that the effect of the self and mutual induction oppose 
each other. 

Alternatively a hummer may be used as a source of alternating 
E.M.F., instead of making and breaking the battery circuit, 
and a telephone used as detector. 

If the self-inductance is less than the mutual inductance a 
modification of the method above may be used. In this case 
the battery and the resistance, r, are interchanged, i.e. M is in 
series with r. 

For balance, the effect of the induced E.M.F. in the arm, AB, 
due to L and M, must neutralize each other. As before, equating 
the numerical values of such E.M.F.'s we have, if c 8 is the current 
through r and M, 

T dc i _m^ 3 

Lc t = Mc 3 , and c 1 {r l + r 2 ) = c 3 (r -J- x), 

M c, r + x 
i.e. — - = - 1 = — ^— ~, 

where x is the resistance of the coil, M. This method, however, 
cannot be performed using alternating current and telephone. 

Determination of Mutual Inductance by the Fluxmeter 

The mutual inductance of two coils may be very readily 
obtained, using a fluxmeter (which is described on page 482). 

AJ } R 

— Mttl'l ' 

Fig. 363 

One coil, M, is connected directly to the fluxmeter, F, fig. 363. 
The other coil, L, is connected through a commutator, C, to a 
circuit consisting of an accumulator, E, an ammeter, A, and a 
1 variable resistance, R. 

The current in L is adjusted to some convenient amount, 
c amperes, and then reversed, causing a deflection of x divisions 
in the fluxmeter. 


If each division of the fluxmeter (as usual) corresponds to 10000 
maxwells, the change in the number of maxwells in the secondary, 
M, due to a reversal of c amperes in L is ioooo* , i.e. for — E.M. 

units of current in L there are 5000* lines of magnetic force 
threading M 

Hence, since the coefficient of mutual inductance is denned 
as the number of lines in the one coil for unit current in the 
other, it has a value, 

50000* . 

for the coils used. 
The experiment is repeated using various values of c, and the 

mean value of - is obtained, and hence the coemcient of mutual 

inductance in E.M. units may be calculated. 

The mutual inductance in henrys is io~ 9 times the above 

value, for two coils are said to have a coemcient of mutual 

inductance of 1 henry when a current change of 1 ampere per 

second in the primary causes an E.M.F. of 1 volt in the secondary, 

i.e. a flux of io 8 lines for a current of 10- 1 E.M.U., i.e. the henry 

is io 9 E.M. units. 



A Modern development of the Kelvin Quadrant Electrometer 
is seen in fig. 364, and is due to Dr. F. Dolezalek. The four 
quadrants, QQ, etc. are supported on ambroid pillars, AA. 
As shown in the figure, two quadrants are mounted on a pivot, 
and may be swung aside to allow of the introduction of the 
' needle.' Alternate quadrants are joined together to terminals 

Fig. 364 

under the base plate of the instrument, care being taken that 
such terminals are very well insulated from the case. 

The needle may be either a light paper frame (fig. 365), coated 
with a metal to make it conducting, or as in fig. 366, a thin 



mica sheet spluttered with silver. In some of the more recent 
instruments a very thin aluminium needle of the shape of the 
second form is used. The needle is attached to a light rod 
which carries a small mirror, M, and is supported by means of 
a thin quartz or phosphor-bronze strip from the torsion head, R. 

Fig. 365 Fig. 366 

The method of observing the deflection produced by a difference 
of potential on the quadrant consists of the use of the usual lamp 
and scale, as with galvanometers. 

A beam of light from a lamp is directed on the mirror, M, 
which reflects the light on to a scale placed one metre away. 
If the mirror is concave, and of the correct focal length, a clearly 
denned image of the source will be obtained on the scale. If 
M is a plane mirror, a lens is required to produce a sharp image. 
Such a lens is placed in the path of the incident beam from a 
source of light below the scale ; the reflected beam does not, 
therefore, pass through the lens. 

The Suspension 

(1) Quartz Fibre 

If a quartz fibre is used as a suspension, the needle may 
be charged by means of the charging device shown at K. 
K is connected to the source of potential and the metal rod 
turned until contact is made between it and the rod which 
supports the mirror on the needle. K is then turned to its 
original position leaving the needle charged. Due to the very 
good insulating property of the quartz, this charge will be main- 
tained. However, there is a danger of breaking the quartz 
suspension during this process. 

For most purposes it is more convenient to avoid using the 
charging device To make this possible the suspension is coated 
lightly with a calcium chloride solution. The hygroscopic 
properties of the latter ensures a conductivity which is sufficiently 
good to maintain the needle at the potential of the source which 
is permanently connected to R. 

An alternative method is to splutter the quartz with silver, 
but this cannot as a rule be carried out conveniently in the 


Suitable quartz fibres are very simply made by the aid of 
a coal gas-oxygen flame. A burner for this mixture is shown in 
fig. 367. A piece of quartz rod is heated in the flame until 
thoroughly soft ; when at this stage the ends are drawn apart. 
If this is done rapidly a fine quartz thread will result. If a still 
finer fibre is required draw out the quartz rod to about 1 mm. 
diameter, and then reintroduce into the flame. When the 
quartz becomes very soft the pressure in the flame itself will 
blow the quartz outwards into very fine fibres. It is advisable 
to have a black velvet cloth on the bench to receive these threads. 



Fig. 367 

Small hooks may be fastened to the ends of the fibre, when cut 
to the correct length, by means of a small globule of shellac 
(solution in methylated spirits) ; a hot iron is held over the 
globule, which is placed on the hook over the end of the quartz. 
The spirit evaporates and the fibre adheres. 

Another method of fastening the hook to the fibre is to use 
indian ink. The end of the hook is dipped into indian ink 
which is allowed to become ' thick ' by evaporating. The fibre 
is then placed on this plastic drop ; when dry the two will be 
found to be very firmly held. 

(2) Phosphor-bronze Strip 

Of the metal suspensions phosphor-bronze strip is most 
satisfactory Platinum and tungsten wires may be obtained 
with a smaller diameter, but usually have the disadvantage 
that the zero of the needle is not stable when very fine suspen- 
sions of these metals are used. 


For the experiments described in this chapter, the finer 
phosphor-bronze strip obtainable commercially gives a sensitivity 
which satisfies the needs of the experiment. If, however, the 
phosphor-bronze strip available is too thick for the purpose, it 
may be treated as described below ; which method enables a 
fair sensitivity to be attained. 

A solution of one part nitric acid to four parts of water is 
taken, and the Suspension immersed in it. The action of the 
acid causes an evolution of gas which adheres to the suspension, 
causing it in time to rise to the. surface. The chemical action 
should be slow. If violent action takes place the solution should 
be further diluted. Of course, a preliminary test should be 
made with a small sample of the phosphor-bronze to be used in 
order to avoid undue waste. 

The strip should be very well washed and dried, and its sensi- 
tivity tested in the instrument. The process is repeated until 
the required sensitivity is obtained. When this process has 
once been performed with the specimen of strip available, the 
time of immersion in the acid required to reduce cross-section 
to a suitable value may be very readily estimated, and the time 
spent in testing will be reduced, e.g. the strip may be sufficiently 
reduced by immersing four times in the acid, allowing it to come 
to the surface each time. 

In this way a sensitivity of 700 to 1000 cms. per volt may be 
obtained with the instrument shown, but for the experiment in 
this chapter a sensitivity of about 25 cms. per volt will be found 
to be quite sufficient. 

The above process is carried out using the wire without hooks. 
If the hooks are soldered on to the ends, the action of the acid 
will sever them. The small hooks are then soldered to the ends 
of the strip, using a very small pointed iron. Care should be 
taken that the ends and hooks are clean and the iron hot. Use 
soft solder and a trace of fluxite. 


To prepare the instrument for use, the following adjustments 
must be made. Both pairs of quadrants are connected to earth, 
and are therefore at the same potential ; R is also earth connected. 
Thus, the only forces acting on the needle are those due to the 
torsion of the suspension. The needle is raised or lowered by 
means of the screw, T, until it swings about the mid-plane within 
the quadrants. By means of the levelling screws, B, the whole 
instrument is then levelled, so that the suspension hangs centrally 
within the quadrants. To ensure this, the rod which supports 
the mirror, M, is sighted along the two diagonal spaces between 


the four quadrants and a slight adjustment made if required to 
bring the rod truly central. The torsion head, E, is then turned 
until the needle appears symmetrical with respect to the quadrant. 
A suitable high potential of, say, ioo volts is applied to the 
needle (in the manner described later, fig. 368). If it were 
adjusted to be precisely symmetrical no movement should result 
on making this change. Any error in the adjustment by eye for 
symmetry may now be corrected, by noting the direction of 
deflection of the needle. The potential is removed and the 
needle earthed. The torsion head is given a very slight turn 
in the direction of the movement previously observed. The 
needle is again charged and the process repeated, until on 
charging and earthing the needle alternately the deflection 
produced is of the order of a few cms. only. To reduce this 
small deflection to zero a slight adjustment of a levelling screw 
will be found su