]\gt ; On riables in a Multiple Integral .
By .
W. H. YOUNG , Sc. D. , F.R.S. ( Received March 25 , 1919 .
) The which enables us to transform th independent variables in a is well known .
It is where we have for ) written for the same symbol repeated times , ( for for the result of substituting in for the 's the which they are of the variables , , , for the Jacobian of the with respect to the for , and for , while the field of integration on the right being supposed to be any region or set of points in the flat space defined by the variables , that on the left is the corresponding region or set of points in the space defined by the variables The conditions under this fundamentally important formula holds have not , however , been stated with any sort of generality , except in the one-dimensional case .
has been undoubtedly due to the very real difficulties which have been felt in the ) one dilncnsion to a higher space .
Neither the ument employed in the } ) problem , nor the conditions obtained , are eneralisable as they stand .
I recently had occasion to form ) a very general set of conditions under which the formula holds in any number of dimensions , provided only the function is ) by unity and the field of integration on the right-hand side is an -interval , a term which it is convenient to use for the eneralisation in ' rectangular parallelepiped whose faces parallel to the three co-ordinate These conditions are , indeed , that it arcely seems possible for them to be improved upon materially .
I propose in prcsent communication to show how , when once this npler case has been dealt with , in which is replaced by unity and the field of integration is estricted to an -interval , the more general formula , stated at the } be deduced .
I have purposely so worked the to prove as to make its value independent of the greater or less ality of the conditions obtained in the simpler problem .
of the dependent Variables in Multiple .
83 S2 .
The theorem is as follows:\mdash ; If .
, , define a correspondence between the x- and the -spaces which satisfies the following conditions:\mdash ; ( 1 ) The functions which the 's are of the 's are continuous with respect to the semble of the latter variables .
( 2 ) Every -interval in the -space has for a definite portion of the -space , whose boundary is the generalisation for snch space of the concept of a simple Jordan curve without loops , so that the boundary divides the whole of the -space into two distinct portions , one inside and the other outside .
( 3 ) The hyperyolume of the portions of -space thus enclosed , and which may in articular be evanescent , are given by the absolutely yent integral taken over the -intervals in the -space , of which they are the images , tho hypervolumes being accordingly positive , negative , or , according as the value of this integral is positive , negative , or zero .
, with the notation explained above , whatever function may be of the variables the field of integration on the any m-interval , and that on the left the space A internal to the boundary corresponds to the boundary the -interyal , p1ovided only the function is summable , not only ovel this portion of the -space , but also over the whole of the set constituted by ) points in the -space , which are the images of all the points of the -interval .
Further , if be any set of points whatever in the x-space the set of points iu the -space which has for image , then provided only the integral on the left exists .
S3 .
The proof of the is the same for any number of dimensions , except as regards notation and terminology .
I propose , therefore , for brcvity Prof. W. H. young .
On Change of the and simplicity of explanation , to assume henceforth that there are only two independent variables , and I shall call the old variables , and the new variables , so that the formula to be proved becomes The method I adopt , a method I have already had occasion to refel to in previous communications to the Society , is that of monotone sequences .
The form in which it is convenient to apply this method on the present occasiom is not that in which the theorem to be proved is first discussed for a function which has a finite number of constant values over a finite number of rectangles in the -plane .
The convenient plan to employ is now to first prove our theorem , when the function is a continuous function of , and therefore also of , while the field of integration on the right is the rectangle Divide this rectangle into sub-rectangles by lines drawn parallel to the axes and so near to one another ) the image in the plane of every sub- rectangle has a span less than , where is itself so small that the oscillation of in each of the little area-images is less than .
This is possible owing to the continuity of the correspondence and the continuity of the function , by virtue of the well-known theorem that continuity involves uniform continuity .
We may , evidently without loss of generality , assume that the area A in the -plane corres o to the fundamental rectangle in the -plane is positive .
We may also suppose the function to be never negative , for , if not , it could be expressed as the diffel'ence of two functions each of which is continuous and neither of which is negative .
Let us agree to denote by the same letter name and the area of each portion of either plane considered .
Let be the typical little rectangle in the -plane and its image in the -plane .
Then the areas are all positive , but the areas are general some positive and some negative .
The areas add up as they stand to the fundamental rectangle .
The areas overla } ) and have for their algebraical sum the area A. Also , as the boumdary of each divides the plane into two distinct parts , they cut one another up into a finite number of non-overlapping areas each of which will necessarily occur an odd number of times , , say , times positively and times negatively , whenever is inside , where varies , of course , in general from one to another .
Such however , as are outside A will each occur even number of times , say times positively and times negatively .
in a le Integral .
We are thus finally left with a set of essentially positive , of which each occurs only inside and does not occur at all outside , their sum being exactly equal to A. Moreover , as every is contained in a , the oscillation of in every is less than .
Hence , by the definition of a double integral of a continuous function taken over the area , differs from the sum of the products of each into the value , say , of at any convenient point in it varying , of course , in general from one to another ) by a quantity which is less than the product of into the sum of the , that is by less than All turns now on expressing this sumrqation terms of two summations involving , which shall call respectively the upper and lower -summations .
By the upper -summation we shall mean that obtained by summing the products of each into a value of , chosen to be the upper bound of in that , when is positive , and the lower bound , when is aCive .
By the lower summation .
shall understand that obtained from the upper summation when vords upper bound and bound are interchanged .
Then we assert that\mdash ; ( 1 ) The -summation has a value intermediate between those of the upper and lower -summations .
( 2 ) The value of the upper -summation exceeds that of the lower by a quantity which is less than the product of into the sum of the , each taken positively .
To prove ( 1 ) , we remark that will virtually occur in terms of the upper summation , as being part of , and will contribute to the upper summation in all , where , , are tlpper bounds of in the positive which contain and , are bounds of in the negative 's which contain of , and each of the 's is .
Hence the contribution of is greater or equal to .
that is , greater or equal to Hence , the portion of the upper summation due to such a reater or equal to the corresponding part of the -summation .
We have also to consider the which occur the number of times Prof. W. H. Young .
Prof. W. H. Young .
positively and fatively outside A. Here the 's of the positive being greater or equal to any value of chosen in are greater or equal to the 's in the negative \ldquo ; .
Hence , though the 's are now equal to the 's in number , the excess of the sum of the former of that of the latter is necessarily positive , so that the portion of the upper summation due to the parts of the outside is certainly positive , while there is no correspondportion in the -summation .
It at once follows that the -summation is not greater than the upper summation .
Similarly , it is not less than the lower summation .
Hence our assertion ( 1 ) is true .
As regards our assertion ( 2 ) , it is sufficient to remark that it is an immediate consequence of our mode of forming the summations , coupled with the uniformity of the continuity of It follows , since the is numerically equal to taken over , and therefore numerically taken over , that the and summaGions all differ from one another by a quantity which is numerically less than taken over the area A. But the -summation has been shown to differ from ( 1 ) taken over A by less than Hence must differ from either of the -summations by a quantity which is numerically less than , ( 2 ) where the integral is taken over A. Now take either , say the upper , of the -summations .
It may be written in the form , ( 3 ) where the integral is taken over , and is the point of at which assumes its upper bound in .
This summation differs therefore from Independent Variables in Multiple lntegral .
by taken over the whole area changing of course by jumps as we move from one of the fundamental rectangle to another .
Now the coefficient of in the integrant being numerically less than , it follows at once that the difference between the double summation ( 3 ) and the double integral ( 4 ) is numerically less than Hence the double integral ( 1 ) differs from the double integral ( 4 ) by less than the sum of the last expression and ( 2 ) .
That is by less than Hence , as is at our disposal , ( 1 ) and ( 4 ) must be equal .
S4 .
Next let denote any bounded function obtainable from the continuous function by monotone sequences , in other words , any bounded lower or upper semi-continuous function , or any lu , function .
The truth of our theorem in this case follows immediately .
In fact , if , for example , , , .
represent any monotone sequence of functions for each of which the theorem has been proved to be true , while the limit of the sequence is a bounded function , then the integral of over the area A is the limit of that of .
Moreover , the monotone sequence of functions of , of which the typical member is , may , by a known theorem , be integrated term-by-term after being multiplied by the summable function .
Whence it follows that , since the theorem holds for every , it holds for .
Thus the theorem holds for every bounded function defined by sequences , or , as we may say , definable mathematioally , the field of integration on the right being the fundamental rectangle , and that on the left the area A. S5 .
We shall now find it convenient to prove the ] theorem:\mdash ; THEOREM.\mdash ; If the conditions of and S 2 for the correspondence between the and the -planes hold , then every set of zqro content in the -plane is either the image of a set of zero content in the plane or the set of positive content of which it is the image , is such that is zero at all points of this set of positive content except those belonging to , at most , a set of zero content .
PROOF.\mdash ; Every set of positive content is the sum of a set of zero content and of an outer limiting set , or as we call it an -set .
It is therefore clearly sufficient to prove that if a set of zero content in the area is the image Prof. W. H. Young .
On Change of the of an 0-set in the rectangle , " " almost everywhere\ldquo ; in this -set .
Let be the subset of this 0-set at which , and let be the of , so that has zero content .
Then the metion which is unity at every point of and zero elsewhere , and the function which is unity at every point of and zero elsewhere , are both necessarily obtainable by monotone sequences .
Writing for in S4 the former function , and therefore for the latter function , we obtain bhen Hence the Jacobian must vanish " " almost everywhere\ldquo ; in ; in othel words , in that part of the -set at which the Jacobian is known to be not negative .
Similarly , it vanishes 'almost everywhere\ldquo ; in that part at which it is ative .
Hence it vanishes " " ahnost everywhere\ldquo ; in the -set , as was to be proved .
S6 .
Now let be any bounded function possessing a Lebesgne integral .
Then we know that it can be closed between a function and a lu function having the same integral .
Also , by S4 , we know ' where by definition Now differs from only as a set of content zero in the plane , and therefore , by the theorem just proved , and either agree for all values of and except such as belong to a set of content zero of the -plane , or , if they differ at other points , , is zero .
Thus we shall make no mistake if , in rating with respect to , we write .
In other words , we have whatever bounded function may be .
S7 .
Now let denote any unbounded summable function supposed , as may be done without loss of generality , to be positive ; let be the function which is equal to wherever , and have the value elsewhere , so that endent Vin Multiple Then it is evident that , by the preceding section , as is bounded , we may write for in the equation at the end of that section .
Moreover , by definition Hence If now never changes sign , so that without loss of gen erality we may suppose it to be positive , the integrant of the integral on the lefthand side traces out a monotone increasing sequence , we may introduce the limit inside the integral , so that if nowhere assumes an infinite value , the required result at once follows .
If , on the other hand , assumes infinite values , we cannot go further unless we llake the convention that is to be regarded as having the value zero whenever is zero .
With this convention , the result again follows by virtue of the theorem oved in S5 .
S8 .
Tf , however , does not preserve the same hout the fundamental rectangle , the above argument fails .
In order to understand the reasoning required in this latter case , we have to bear carefully in nlind the sense in which we have used the word image .
The point which corresponds to an assigned point of the fundamental rectangle in the correspondence is not necessarily a point of image A of the fundamental rectangle , but may be outside in the gregate such points , as will be seen by referring to S3 , belong to areas which cross one another out , so that the points themselves may be said , in a certain sense , to cross one another out also .
The employed in that section shows that we require to assume that exists and possesses the properties ascribed to it , uot only over the image A of the fundamental rectangle , but over all that portion of the plane outside A constituted by the points which correspond t , o points of the fundamental rectangle without being included in A. It is the recognition of this tacitly assumed existence of outside which enables us to complete our reasoning .
value , in fact , of at a point of the fundamental rectangle which has for image a point outside , is necessarily the values of at this latter point .
us then now define a new function equal to at all the points , both inside and outside , which correS } ) to the points inside the fundamental rectangle where is positive , and equal to with its sign changed at all the points , both inside and outside , which correspond to points of the fundamental rectangle where is 90 Change of the Independent Variables in Multiple Integral .
Then it is evident that ] , as is summable not only inside but outside , so is , is everywhere positive hout the whole of the .
Also let be a function which is defined to wherever , and let be to where ositive and greater than , and be equal to , wherever negative and less than .
Then , since is bounded , it follows by means of S6 , in the same manner as in S7 , that ? ?
Also the integrant of the integral on the left is clearly positive .
Hence , a reasoning isely aogons to that of the latter part of S7 shows that Thus the integral on the left-hand side , taken over the whole of the fundamental rectangle , necessarily exists .
is everywhere numerically positive function and is accordingly summable over the lndanlental rectangle .
Hence neccssarily 1lnmerically less than this latter function , that is , the succession of which it is the typical term has every one of its terms numerically less than one and the same summable function of .
Hence , by the theory of of sequences , we may integrate it term by term , and and , with the sanJe conventio ] ] as that explained in S7 , This last expression is equal It remains only to replace the area , which is the field of ration of this latter integral , by any set of , in that area , and the fundamental rectangle of the -plane by the set which is the image .
To show that this is possible , it suffices to remark that we may replace On Certain Independent Factors in Mental Measurements .
91 function defined product into that function of which has the value unity at all points of and is zero elsewhere , and , accordingly , , by the product of this function into a function of , which is defined to have the value unity at all points of the set which has for image and is zero elsewhere .
We then at once have the desired result , except that there remains a doubt as to the type of summability required for .
This must be such that , when multiplied by the auxiliary function just introduced , it is summable over the whole set of points which corresponds to the fundamental rectangle .
But this is clearly the same as demanding that should be summable over the set of points constituted by all the points in the -plane which correspond to the set , that is , over the set Hence our theorem has been proved , with all the generality given to it , in the enunciation of S 2 .
On Independent Mental Measurements .
By J. C. MAXWELL GARNIBTT late Fellow Trinity College , .
( Communicated by Prof. A. N. Whitehead , F.R.S. Received March 26 , 1918 .
) I. In a paper published in the ' American Journal of Psychology ' for 1904 , Prof. Spearman argued that " " all branches of intellectual activity have in common one fundamental function or group of functions ) even that this fundamental function enters into sensory discrimination no less truly than into more complicated intellectual activities .
In 1909 Mr. Cyril Burt experiments on two groups of Oxford schoolboys undertaken " " with a view to testing in practice the mathematical methods of Prof. Spearman Mr. Burt calculated the correlations between the performances of his subjects in 12 different tests , which he classified as sensory , motor , sensori-motor , and association tests , together with one ( the twelfth ) test of voluntary attention .
Mr. Burt 's results were consistent with the view that all the correlations were due to the operation of one and only one eneral factor .
Two years later Prof. Spearman and Dr. proposed a new test 'American Journal of Psychology , ' vo ] .
15 , p. 284 ( 1904 ) .
' British Journal of Psycholog vol. 3 , pp. .
'British Journal of Psychology , ' vol. 5 , ) .
VOL XCVI.\mdash ; A.

