]\gt ; Th Vector Properties ofA lternating Current it is seen that we have meaos at our to separate any truel iodicity of a variable among its ular changes , can extend the time limits sufliciently .
The proof of this ppition lies outside the limits of this papel The applictio]l of the of to the ve.tiou of may be called " " hiddell\ldquo ; periodicities , all lnstance of be extended , and ults are obtained a number of periodicities , such as those ) depcnd on the rotation of the sun bout its axis , al.c examined .
A full treatment of the subiecb will Magnetism .
: : ' The ) perties of Altel .
and Periodic \ldquo ; By OMPNER , ) .
Cornmunicated O. HEVIICI , \mdash ; Read Junc 17 , 189 Il for thnt iatious of a silnple hmonic f , such :us A , can be represcnted 1ou the , on a iixed , of vectol .
of constnnt ] t posed to evolve .
so as to complete ono olution : time by the ntlon ) The , vector and the fixed line is at any instant .
In the first cdition of Thomboll and 's ' tural Philosophy ' ( vol. 1 , p. , it is shovvu that auy simple monic functions of period can be compounded to a simple harmonic funclion of silme period , and that the vcctor , compounded Ltnction , is ined from thosc epresenting the component of vector addition .
This proved useful many , but it ' proved especially collnexion with altcrnate rrcnt problems .
Its } to cases lvas first clearly .
T. H. kesley l lwve .
The of it rent work devcloped into may bc cribed as avector , or , method of enti alternate current quftll lcb .
method the of the denotes the of the curl.cnt or any eprcsents time which eJapsesbetweell the tants a which maximm of the qunntilies .
In rbicular , if one represe the on conductor , and a second denotes the strength of curJent in it by oltage , power absorbed in the conducto is measured by th product of length of two lines into ' cosinc of the angle veen them .
is to ; possibility of Dr. W. E. Sumpner .
The Vector Properties of deduciug from the network figure the electrical power absorbed by the conductors , as well as the magnitude of each current , and voltage , that is chiefly due the usefulness and importance of this mode of representation .
It will be convenient here to describe brieiiy the mathematical basis on which the method rests .
It will be evident that we may represent the value of a simple harmonic function A either by the projection of a revolving vector on a fixed line , or by the projection of a stationary vector on a revolving line .
It will be more convenient for our purpose to adopt the ]abter plan .
If , therefore , we have a vector , of length , the projection of this on a line supposed to revolve uniformly will always be equal to A , provided the period of revolution is such that , and provided also that the line coincides with the vector at the instant from which is measured .
We can , moreover , represent the value of by the projection at the same instant on the same revolving line of a vector of length , provided only the vectors and are perpendicular .
It will also be evident that any other simple harmonic function of the same period will be similarly represented by the projection of the and the length of this vector will be equal to the maximum value of the function it represents .
, in any system of conductors traversed by alternate ents , every electromotive force is a simple harmonic function of one period , , and if the coefficients of induction and the capacities are all constant quantities , it follows mathematically that each curre1J , and each voltage , can be expressed in the general form which can thus be represented by the ector where and are the perpendicular vectors representing and pt .
If and are unit vectors , the length of is , which is the maximum value of the function F. The magnitude of an alternate current quantity is however estimated by the square root of its mean square , and not ) its maximum value , and since Alternating Currents and other Periodic Quantities .
467 if we choose for and two perpendicular vectol .
S of lengbhs numerically equal to 1 the length of will be equal to , or to the nagnitude of F. now we consider the value of the mean product of functions , we see that this mean product is or the same as the scalar prodnct of tho ' vectors , .
Every alternating current problem is reducible to the consideration of the curront , voltage , electrical power spent in the diff . .
Since all these quantilies are periodically some value has to be aken .
and voltagc the mode of ) selected is vays that obtained by taking the square root of the mean value of the square of the quantity the period .
The electrical power is , vevel .
, the mean through the period of the product of the values voltage and current .
Hetlce , if the current and voltage are represented by vectors in the manner described , tho scalnr of { he two vectors csent the .
This vocto ] .
method particularly suited to curl G oblcn ] ) been very much such ) ) oses by , hllcc it to understand and 1 npply , and it catl be used for the simple solution of mally p hich ould o ncetl tl use of the calculus .
The method , ever , ssumcs t th cs currcnts considered vary : ) ) illc 1 mercial practice is ) the ) icfIy to saturation and cnomcn in the ) , and also to the fact namos do not devclop force varying ling to a .
The object of the present ] ntion is to far ) lcthod C be modified , and ectly , in cases wllcrc the , although periodic , do not vary to ) simplc we shall cquently to to the of root of the of periodic fullctiolJ , shadl call such a Dr. W. E. Sumpner .
The of quantity the magnitude of the periodic .
Thus , if is a function of the time , periodically repeating itself after every interval , themagnitude of , which we shall denote by , is given by the equation magnitude of When we are considering two such functions , and , the mean value of their product will be denoted by The mode of measuring electrical power known as the Three Voltmeter Method , really dependent upon a curious property of periodic functions which may be stated as follows:\mdash ; If any arbitrary periodic junctions , and , be repl'esented in magnitude by the lengths of two sides of a triangle , and if the side represents the magnitude of ( another periodic function equal to the sum : or difference , of the former two ) , then the mean value of the product of any two of these functions will be represented by the prodnct of the lengths of two corresponding sides and the cosinc of the ] between them .
Thus , in the figure , if epresents v , and BC or represents then BA represent , and DA will represent ; also the mean product of be ABC , the mean product will be CAD , and so on .
Thus we regard and as vectors , the length of the vector representing magnitude of the periodic function , and the scalar product of vectors tho mean product of the two corresponding functions .
periodic function derivable from and , by a linear relation , can be similarly represemted in with the ordinary propel'bles of vectors , and the scalar product of any two such vectors will be } to the mean producb of the corresponding periodic functions .
In order to establish bhis we may proceed as follows:\mdash ; The Me surement of the Power given by any Electric Current to Circuit Professor W. E. Ayrton and Dr. W. E. Sumpner , 'Roy .
Soc. Apri19 , 1891 .
lternating Currents and other Periodic Quantities .
Whatcvel be the law of variation of , and , if . . . . . . . .
( 1 ) , then , taking means , we have hich eqnation is the snme as But if the lengths of the three sides of triangle respectively reprcsent , and , we have geometry herc is itnd Comparing ( 2 ) ( 3 ) , it will bc that for l'cprcsented by the line AB the positive taken in cquation , while for other , he ativcigns must ) chosen .
In either we co.s lIoreove ] .
, if in equation ( 1 ) ) ither t , and occed as before , we find two sicles of the ABC be chosen , hc products of these sides into the coslne of the anglc bebwecn them is equal the lllcan product of the .
periodic functions .
oing tment t ; that to to the ) of ftnd sucl , : propertics ated .
But we yet to ) it is ossible .
to , or , in .
vords , tl at of theso q titics arc if y the mc.th shall therefol.c ) that ovided can establish inoquality : 470 Dr. W. E. Sumpner .
The Vector Properties of so that if we can show for two periodic functions that the product of their magnitudes is necessarily greater than the mean value of their product , we shall have shown that it is always possible to construct a triangle the sides of which represent the magnitudes of any three periodic functions related as in equation 1 ) .
The of ( 4 ) can be illustrated by supposing that repreents an alternating voltage applied to a conductor , and producing a , through it .
The right-hand element of ( 4 ) then represents the power :bsorbed , whils the left-hand elemen{ , is the product of volts and amperes .
The fact that the power absorbed in an inductive circuit is always found to be less than the product of volts and amperes , is usually stated by asserbing that the " " power factor\ldquo ; is less than unity for such circuits .
xperience has not shown any case in which the power factor of a circuit is greater than unity , but this of course is no evidence of the trulh of ( 4 ) , which must be proved for the general case in which the two periodic functions are perfectly arbitrary Fourier series .
* Let us therefore put . . . .
( 5 ) . . .
( 6 ) .
By taking the mean value of twice the square of each of these equations , we get , , and by taking four times the value of mean producb we get , the proof given , the two functions and are assumed to have fundamental periods that are commensurable with each other .
This will alw .
be so in practical cases , and what happens with functions of able .
periods is of but theoretical interest .
It is , however , vorth notice that no limitation is thereby imposed on the application of the theorems given in this paper .
These theorems deal with mean products , and the of them 18 dircctly deducible from the inequality given in ( 4 ) .
Now when ] periods are the prodtlct consists of terms such as npt , where and incommensurabie , and as each of these can be expressed as the sum of sinuous functions of angles denoted by , the mean value is zero , the vectors representing and are perpendicular , and the truth of ( 4 ) is at once evident .
But smuch as the value of may be infinitesimally small , it may be to take infinite time to determine the me valucs .
and titiev .
and we hall have oved ( if we ] that , or at This will be if we can establisl ] the ebraic i it be found easy this by an inductive method , but there is no need to do so here , s1nce it is in the t of detel.minants that and is cessarily positive .
follows that the i ality ( 7 ) is trne .
that the inequdity]nay become an the necessary and fficient conditions are I ords , the two must a each at every installt of time , and muht be of onriel .
xnction .
This is a limiting is llot in ( 4 ) , with ihis exception only , } er uoticed : \mdash ; .
and ) be , } .
consfanls by ) B3 Cssion i we get , .
is mennt ) ) of .
, , we obtnin by ) king ) } ) ) 1 , Dr. W. .
Sumpner .
The Vector Properties of The product of the magnitudes of two periodic functions always greater than the mean lfalue of product .
From this and from the foregoing reasoning , it follows that : Any two periodic functions be represented vectors in such a way that tloe length of each vector r.epresents the magnitude the and the scalar product of the vectors , the mean product of the funchons .
Any other function derived from , the first two by means of a linear relation can be represented in magnitude by a vector derived by means of the linear relation from the two original vectors .
The scalar of any two such vectors will be equal to the mean product of the sponding fu nctions .
Now suppose and are two vectors representing two arbitrary Fourier series , and .
It follows from the law of vectors that any vector in the plane will be a linear function of and , and must therefore represent a Fourier series derivable from and in a linear manner .
If , therefore , we a third arbitrary Fonrier series , , which cannot be expressed as a linear function of and , it follows that his cannot be represented by a vector i same plane as and .
It remains to show that a vector , , out of this plane can still be found to represent Let us first of all assume that three vectors ( not in one plane ) have been found to represent three Fourier , and .
We can then show that any linear function of , similarly represents the Fourier series , which is the same linear function of For if , in accordance with the notation already used , we denote by the symbols the square root of the mean square of the , i.e. , the magnitude of the mean value of the product the length of the vector the scalar product of the vectors If we substitute from the above relations , we get where represents the value of It follows that is always less than , except in the limiting , in hich is zero at all parts of the pel.iod .
This will only occur if at every instant , when \mdash ; June 11 .
1897 .
] Aliernating Currents and other Periodic antities .
then have If , now , we compare the value of the square of any vector with the value of the mean square of the corresp .
Fourier series , we find , by expanding and using the above relations , that the expressions are identical ; and , moreover , it is clear that the scalar product of any two such vectors is identical with the mean value of the product of the corresponding pair of Fourier series .
We have now : however , to show that whatever may be , it is always possible to find three vectors , , to rep.resent them .
For this purpose Jet us consider two other .
Fourier funct ons , and , derived in a linear manner from the glven functions , , in accordance wibh the equations and where , and are scalars determined from the mean products , It will then be seen that and and whatever , and be there is a real angle , , for which since the product must be less than If thereforo , we take two planes inclined to each other at an ang e , , and take two vectors , , in these planes , drawn perpendicular to the line of intersection , and of such length that and and and if , in addition , we choose a vector , , alollg the line of intersection of these planes , and of such length that Dr. W. E. Sumpner .
he for Properties of we shall then have the three vectors completely represent the series Since linear functions of , and , while are the same linear functions of , and , it follows from the foregoing reasoning that also represent the periodic functions It is thus established that.\mdash ; Any threeperiodic functions can be rep } esented by vectors in such a way that the length of the vector represents the magnitude of the function , while the scalar product of any two vectors represents the value of the product of the corresponding functions .
Any other periodic function derivable from the three , in a finear manner , be ilarly represented by a vector , derivablefrom the original three vectors by me the same linear relation .
It will , however , be seen that it is impossible , in three-dimensioned space , to find vectors represent four arbitlary periodic functions , unless a linear relation connects them .
For instRnce , the functions pt , cannot be so represented .
The mean product of any two of is zero , and it is impossible in threedimensioned space to find four lines each of which is at right angles to the other three .
In discussing problems the solutions of which are expressible in terms of the mean square of a Fourier function , or the mean product of two such functions , it may sometimes prove advantageous to make use of the foregoing theorem .
As a rule the complicated Fourier series ocourring in problems obtained in an indirect and somewhat artificial manner by expressing in mathematical language the value of some periodic quantity which can be represented graphically in a very simple way .
In some cases the curve , though perfectly determinate , follows no recognised malhematical law ; in other cases it consists of a series of discontinuous culves , each of simple mathematical character .
In all such instances the mean squal.e of the ordinate of such a curve , and the mean of the esponding ordinates of two such curves , can more easily and quickly be obtained than the Fourier expansions themselves .
If the problem only demands a knowledge of these mearl squares and mean products , the vector method wiIl probably be the simplest } other Periodic Quantities .
475 to and possibly many exist which may be solved in this mannel .
This seems the more likely when it is remembered tl at , in problems involving periodic forces aJtd fluxes , the depends ) ( of the force and its ) espon or , if the forces and fluxes , upon the mean square of the of .
The theol.em , owever , seems nlost applicable to the case of lternate crent problems .
To apply th to to ents let us first of all consider case of a lletwol.k- of conductors in which are only tral .
iodic elecbromotive forces , , and the bl.anches .
esistances are all non-inductive , it -ill -ays then be possible to express the current eac branch , and the ally 1 oints as linear function , and coefficicnts only the resistances of the network- branches .
If , and al.e the periodic functions , it will be possible to reprebent all the and all urrellls involved by means lletvork of vectors .
A case in which of one of the branches is to its ision of a coefficient of self-induction , calt best be ated by that a branch possesses arbit electl.omotive force , related the cul.rent ssing trough the in a manner given by the equation k of elation b the vectors and excel ) their is , and a sequcnce perpendictllar to cnch othc]. .
No 1 , wasted , due to self-induction , since relation ives a rtuin collnexiolt belween the coefJicients hich c linenr function of the denoting the , but doeH n makc it ible to express a linear unction of other electromotive and althongh we , the ] will not be ualltities , since they will ltot ( Illy of the brancl es of the } ) ( .
It will be possible to avectol to ) only special cabes will ectot .
be the snmc ) lane as denoting and Similarly a cnse in , of the uetwork- are coltnected the tel.minals of a .
of capacity can best be treat 476 Dr. W. .
Sumpner .
The Properties of by assuluing that the condenser is replaced by a conductor possessing an arbitl.ary electromotive force , related to the current passing through the branch in a manner given by the equation The vectors representing and are such that their scalar product is zero , and the same remarks apply as in the preceding case .
The case of a transformer can be similarly treated .
The reaction due to the varying induction of the core prodnced by a cnrrent through the primary , must be treated as an arbitraly electromobive force E. The case will differ from that of constant self-induction , in that the vectors representing and are at right angles in the latter case , but not in the former , since , owing to hysteresis and eddy CUl'rents , the core absorbs energy , and the scalar product of the vectors and cannot be zero .
It will thus be seen that the effect of inductance in branch of the network is the same as that of an arbitrary electromotive force placel there , whether the inductance be due to constant , to condenser action , or to transformer action .
With this understanding it follows that a vector figure can be ' space , completely representing the alternating currents and potentials in a network of conduetors , provided the number of independent alternating electromotive forces does not exceed three .
This jigure can be drawn in a plane if the number of such electromotive forces is not more than two .
Of course , where a linear relatioo , with scalar coefficients , connects three eIectromotive forces , only two of these are to be regarded as ) Now the vector method which has hitherto been so largely used in explaining alternating current phenomena , will not be really serviceable in the new and extended application here suggested , except in cases in which the vector figure is either confined to one plane , or can , with all essential accuracy , be treat , ed as if that were tho case .
Fortunately , severaJ circumstances conspire to simplify the application of the theorem to alternate curreut problems .
In the first place the vectors which are needed divisible naturally into two sets : one representing the cnrrents , and the other the tageS .
It is never necessary to find the product of two of these ectors unless they belong to different sets .
Although in some cases it may not be possible to construct the complete vector figure in one plane , it may still be possible to construct two figures\mdash ; one for each .
In practical cases one of these figures ( say the vectors ) will almost certainly be confined to one plane , and if the vectors of the other ( or voltage ) network be projected on to this plane , the product of any projected voltage and the appropriate cllrrent A rnating Currents Qnantities .
477 vector will accurately repl.esent the corl.espollding .
quantity of power , since it is a well-known property of vectors that the scalar product of any two vectors is the same as the scalar product of one of them and the perpendiculal projection of the other on any plane contaimng the first .
It will only rarely happen that a plane figure will fail to give a sufficiently accurate representation of the relations between the volts and a lternate circuits .
These are generally simple character .
Complic tion is not callsed by multitude of conductors unless some of these are inductive .
A work of non-inductive resistances may involve a large number of currents and potentials , but they all connected by linear relations , and ehecoIlstruction of the vectol network presents no difficulty .
Whera transformer is used , the primary and secondary circuits can be treated as quite distinct .
The vector figures representing them can indeed be drawn to scales , the only necessary connex on between them being the relation which connects the powe ] .
put into the core of transformer with that taken from the secondary .
As a type of the most complicated likely to occur , we may take the case of an alternator feeding the primary eoil of a transformer , the primary terminals of which are also ited through a condenser .
Here we have three inductive circu ts parallel .
There are only three curl.ents to consider .
One of these call be expressed as the sum of the othel .
two , and hence the curx.ent vectors lie in a plane .
There is only one .
This may not lie in the plane of the currents , but projection on this plane will give all that is nee ded .
Let be the electromotive force of the dynalllO , the resistance 0 the armature and leads up to the termin of the transformer , kept at let be the armature current , we The ctors r and may not be in the plane of th CUl.reuts , but their projections on this plane are vectors which are c.onnecte with each other by the same linear in the above iu ncal .
lie all the pl.ojectionb may be tsken to reprcsent the The chief )pose f this vector method has erto been is to explain or to predict of the used in current circuits by variations in resistance , \fnof ; g ion , pacity , and to deduce relations which must be fuliilled in that resuits may be obtained .
It has been possible to do this because : network epresenting one it has been easy to draw the nrcs representing cases by illcreasing the of certairl of the line , in oper proportion .
Vector Proertof Alternating Currents , It will be evident that the same process is equally applicable when the currents are no longer of a simple er , provided only that the twork resenting one , particular ci sc can be properly drawn ; for , however complicated the network may become wheu it represents several such cases , it is obta from the origin network by increasing some of the vectors to correspond with the changes in the conductors .
If we have a current , ssing through ( 1 ) a of resistance a coil of self-induction , or ( 3 ) condenser of capacity , the voltages at the ends of these condnctors iven respectively by the expressions and hence representing them are tively to , and and the effect of an alteration in , or can generally be easily indicated on the network figure .
It may , therefore , be taken as established the vector method of representing aIternate current antities i as applicable when the curl'ents and potentials in a manner different from the sine law as when they are simply sinuous .
But it is desirable ill this counexion to repeat and emphasise an old wal.ning about continued use of coefficients of self and mutual induction in the theol'etical investigatious of alternate cul.rent problems .
These coefficients may be of service in connexion wi coils not possessing iron cores , but they are very misleading in commercial cases in which iron is present .
By no choice of the value of these coefficients is it possible to epresent the phenomena of hysteresis .
The presence of the latter implies that the current has a different law of tion from that of the voltage , while the ssumption of constant coefficients uecessitates that if either the or potential a sine law the other must ] , constant seIf-induction involves the assumption that there is no loss of energy in the magnetic changes going on in the iron , an assumption which is not only known to be wrong , but one which involves an errol of serious practical importance .
It has already*been pointed out that the difficulty of theoretical formulae with experimental results does not arise so much from assuming that the currents vary according to a simple sine law , as from the quite unjustifiable assumption and of constant coefficients of induction .
trician , eorv , \ldquo ; by Dr. W.

