PROCEEDINGS loDiiI ^uitii) of §irtoria o' VOL. XVII. (New Series). PART T. Edited under the Ai/t/iority oj the Coinicil. ISSUED SEPTEMBER, 1004 (C'lnfiiiiiiiig Papers redd bef.ire the Society during; tlie ijiantlis nf May, June, July, 1 004). TlIK AUTHORS OK THK SEVERAL l-ArKRS AllE SEVERALLY RKSI'ONSIHLE KOK IIIK SOUNDNESS OK THE OPINIONS GIVEN AND KOR TIIF. ACCURACY OK IIIK STATEMENTS MADE THEREIN. MELBOUKNE : FOKD & SON, PRINTERS, DRUMMOND STREET, CARL'l'ON. AGENTS TO THK SOCIETY: VVILLIAAIS & NORGATE, 14 HENRIETTA STREET, COVENT GARDEN, LONDON. To whom all coniinunications for transmission to the Royal Society of Victoria, from all parts of Europe, should be sent. 1904. CONTENTS OF VOLUME XVIL, Pt. I. PAGE Art. I. — The Alternate Current Transformer. By Thomas R. Ltle, M.A. (Plate I.-V.) ... ... ... 1 II. — Contributions to our Knowledge of the Anatomy of Notoryctes tyi^hlops, Stirling. Parts I. and IE. By Georgina Sweet, D.So. (Plates VI. -IX.) ... 76 III. — The Relations of the Granitic and Lower Palaeozoic- Rocks near Dandenong. By Ian M. Sutherland. (Plate X.) ... ... ... ... ... 112 IV. — The Antiquity of Man in Victoria. By J. W. Gregory, D.Sc, F.R.S. ... ... ... ... ... 120 V. — Revision of the Australian Aphodiides, and Desci-ip- tions of Three New Species allied to them. By Rev. T. Blackburn, B.A. ... ... ... ... 145 VI. — Tabulated List of the Fossil Cheilostomatous Polyzoa in the Victorian Tertiary Deposits. By C. M. Maplestone ... ... ... ... ... 182 VII.— Catalogue of the Marine Shells of Victoria. Purt VIII. By G. B. Pritchard and J. H. Gatliff ... ... 220 VIII. — Fossil Fish Remains from the Tertiaries of Australia. Part I. By F. Chapman, A .L.S., and G. B. Pritchard. (Plates XL, XII.) ... ... ... ... 267 IX. — New or Little-known Victorian Fossils in the National Museum, Melbourne Part IV. — Some Sihirian Os- tracoda and Phyllocarida. By Frederick Chapman, A.L.S., &c. (Plates XIII.-XVIL) ... ... 298 X.— Contributions to the Palaeontology of the Older Terti- ary of Victoria. Gastropoda, Part II. By G. B. Pritchard. (Plates XVIIL, XIX.) ... ... 320 XI. — On Some New Species of Victorian Mollusca, No. 7. By G. B. Pritchard and J. H. Gatliff. (Plate XX.) 338 37821 ^^tii^^z p p. 10 p. 13 p. 13 p. 17 p. 18 p. 28 p. 29 p. 41 p. 42 p. 42 p. 43 p. 47 p. 48 p. 48 p. 54 p. 61 p. 62 p. 65 p. 70 p. 73 p. 73 ERRATA. Art. I. line 24, insert "closed circuit" before "type," and delete "treated in this section." line 22, read "E^/D'" for "Ej/Di." read "Cos(S + <^)" for "Cos(^ + <^)." read "6'^" for "^^." read "§14" for "§ 15." read "Cos<^" for "CosP." and 3, read "P and POosc^^' for " P., and F.^oscfi." read "^" for "." 7, insert "a" after "of." read "mutually" for "naturally." />, />, read " A_j + y^" for " /;, + jjy" read "§ 47" for "§ 45." read "j^V' ^or ":c^." read " ///O./" for " //,C./." read "y^f^i" for "//'.j." , line , line 11, , line 6, , line 19, , line 19, , lines 2 s , line 4, , line 7, , line 19, , line 11, , line 31, , line 8, , line 29, !, line u, , line 29, , line 4, , line 8, , line 11, >, line 8, ), line 30, ), line 10, 5, line 16, 1 -/?.:' P - I „ ^ /' - read " — ■■ — A for " — r I ^ I read "jcV' f*^'' "a^V" read "§ 34" for "§ 33." insert reference number (V). after "in" insert "Section II. and in." read "e " for " e read " P/' for " P." delete " that." ,8." Art. 1. — TJls Alternate Current Transformer. By THOMAS R. LYLE, M.A., Professor of Natural Philosophy in the University of Melboiirne. (With Plates I.-V.). [Bead 9th June, 1904]. The following paper is divided into three Sections. In Section I. the mathematical theory of the closed-circuit transformer for sinusoidal wave forms is developed, and reduced to a form suitable for practical application. In Section II. is given an example of the application of the practical formulae obtained in Section I. to the design of trans- formers to operate different classes of load. Section III. contains analytical investigations relating to magnetic leakage in transformers, to what are called the transformer numerics, and to tlie determination of the most efficient shapes of transformers of different types as well as a general solution of the transformer problem in which no assump- tions with regard to leakage are made. Section I. 1. It is well known that when an alternate magneto-motive force (M.M.F.) operates in a magnetic circuit (laminated), the M.M.F. per unit length (H, say) and the average flux density (B, say) can be expressed as follows : — H = Hi[Sinzf;/ + k,^in3{wi -js) + (fee] B = Bi[Sin(z£// -8) + l>Sin3{wi-fis) + ^c.] where the period is 2TrJ7V ; and that the iron losses per cycle, per unit volume, due to hysteresis and eddy currents are equal in this case to ^i^^[SinS + 34^3Sin3(/33 - y,) + &c.] 4 If Bi = yLioHi then /Aq and 8 will depend on Bj, lu, and the wave form of H, as well as on the quality of the iron and the thickness of the laminae. 2 Proceedings of tli.e Royal Society of Victoria. In some experiments on good transformer iron of thickness .04 cm. (qp-)) I irdve found by means of my wave tracer^ that ju.,, and 8 are given in terms of B, for periods .03 and .06 sec. by the curves shown in Fig. I., where the curves giving the correspond- ing iron losses are also sliown. In these experiments the wave forms of H were peaked, that is, the value of H at the crest was greater than the amplitude H^ of its first harmonic. When the wave form of H is flat-topped, both fXf) and 8 are smaller for the same values of Bj and 2i>. The values of B at the crests of the flux waves, corresponding to different values of Bj the first harmonic, for the period .03 sec. are also given in Fig. 1, by the upper row of figures along the axis of x. When the third and higher harmonics of H and B are neglected, the above equations take the simple forms, M.M.F. Length Flux = HSinyf// = BSin(a^/-S) Section B = /i,II Iron losses jjer cycle HB^. ^ Volume of iron 4 Iron losses per second wW„. _ ^r^, ^-F-- = Q — SinS Volume oi iron cnrfx which relations will be used in the following approximate theory of the transformer. Notation. — The different periodic quantities considered will in the text be represented by letters such as E^, Cj, E^, C„, with bars over them when the conception of both their amplitudes and phases is involved, while the amplitudes of these quantities will be represented by the same letters without the bar. Letters with the number 1 subscribed will refer to the primary, and with the number 2 subscribed to the secondary circuit. The period of the alternations will be ^ttJiv. 2, On the vector diagram. Fig. 2, let OR represent in ampli- tude and phase the resultant flux F looped on both the primary and secondary coils of a transformer. 1 Phil. Mag., Nov., 1903. Alternate Current Transformer. This flux is produced by the auipere-turns w^Ci + WjC,, (a vector sura) so that Arnp.F or F = o-aiiip.(«,Ci + WoCg) where cr = ivr x permeance of the magnetic circuit. . Section of iron „ , , . . = '+7riu ; , tor a closed circuit, Mean Jength of iron and F is behind ;/jCi + WjCo in phase by an angle 8. (See >; 1). [It will be shown that, tliroughout the range of operation of a transformer, when the primary volts and frequency are fixed, F is very nearly constant, S(j that 8 and /x will be very nearly constant. On referring tu Fig. 1, it will be seen that 8 is fairly constant in any case in the neighbourhood of the flux densities generally used in transformers, and though, at the same densities /xis changing rapidly, we shall not, on account of the approximate constancy of F, introduce much error by assuming both 8 and /x constant during the operation of the transformer.] Hence from O draw OM, ahead of OR by the angle 8 = ROM, and in length equal to F/o-. OM fully represents n-^G^ + n.^C^. In addition to the magnetic lines forming the main flux F and looped on both circuits of a transformer, tliere are others, the leakage lines, which are only partially looped on the circuits and whose action must be taken account of. In Section III. of this paper it will be shown that, after the transformer is somewhat loaded, the effect of these leakage lines on its operation is the same as would be produced by two fluxes ; one, the primary leakage flux, in phase with the primary current, and supposed to consist of lines that are looped on the primary circuit and miss the secondary, and the other, the secondary leakage flux, in phase vvith the secondary current and supposed to consist of lines that are looped on the secondary and that miss the primary circuit. Let these two fluxes be specified by x^anS^^ and x^cthXj^ where :Vi and x.^ are what we will call the leakage coefficients of the two circuits, and cr = 47rX permeance of the magnetic circuit as before. 3. The e.m.f. in the secondary coil being equal to ^F 4 Proceedings of the Royal Society of Victoria. is represented by the vector RS equal in length to 2vn„'F and behind F in phase by a right angle. For the present we will assume that the secondary current Cj lags behind the internal secondary e.m.f. E'j or RS by an angle ' and OM = F/o- we iind that ONo,.„A=A-J, ',.^^,...C, - where A"= 1 + 2^'Cos'. If the angle MON be called )^, we tine), by projecting the sides of the triangle OMN on OR and on a line perpendicular to OR, the relations A'Cos(x f 8) = Cos8 + ^'Cosc^'Sinc^' A 'Sin(x + 8) = SinS + ^'Cos- <^' which will be useful. 5. From O along ON cut ofi' a length OB that will re^Dresent rjCj, where r^ is the resistance of the primary coil ; OB will represent therefore the effective e.m.f. that produces current in the primary coil, and will be the vector sum of (a) the impressed e.m.f. El, (^) the e.m.f. 'di _ due to variation of the main flux F, and (r) the e.m.f. Alternate Current Transformer. 5 d — ~ "i^(.^io-«iCi) due to variation of the primary leakage flux. Hence from B draw BC perpendicular to ON and equal to wX'ji^(jQt^=x-{ryr^^ say, wn^cr where t, = CB will fully represent {c). From C draw CE perpendicular to OR and equal to wn-^, EC will fully represent {b). Join OE. OE will fully represent Ej, the e.m.f. impressed on the primary of the transformer. 6. At this place attention may be drawn to the importance that will be attached in what follows to the quantities Tj, Tj, and $'. As Ti which we will call the numeric of the primary circuit of the transformer or, shortly, the primary numeric, is equal to w «/ «]^ «!«! and ; = —r- ''i A pA I' a, where a^ = sectional area of primary wire /j = mean length of primary turns p = specitic resistance of copper also ar — 4:Tr. permeance of magnetic circuit. We see that tj is equal to Attw into the conductance of the primary wires, considered as one turfi or belt, into the permeance of the magtietic circuit. T2 is a similar constant for the secondary circuit, and will generally be nearly equal to tj; we will call it the secondary numeric. In what follows the ratio of Tj to Tj will where necessary be denoted by/ so that Wi" n^a^ /-- = — = -^ On the other hand & is a variable, varying with the load on the transformer, and for a given load-power-factor approximately as the load. 6 Proceedings of the Royal Society of Victoria. R., being the total resistance or its equivalent in the secondary- circuit ^ ^~Rr~R, It is worth noting that Tj, To, and 6' are of zero dimensions. 7. Returning to the diagram Fig. 2, if we call the angle EOB a, so that Cosa is the power factor of the transformer, we find by projecting the figure OECB on ON and on a line perpendicular to ON, that EjCosa = r,C, + K;/?iFSin(8 + x) -= r,Q,[\ + ""' ,Sin(S + x)] = riCi[l + A,(Sin8 + ^'Cos-<^')] making use of the relations in ^ 4 and 5. EiSina = .TiTiriCi + 7e'«iFCos(S + x) = ^i^iixxT^ + ~,Cos(S + x)] = riCi[a-,Ti f ^,[0os8 + ^'Cos'{x,Sincf>' + 9^) + Ti T] ^ Dividing EiSina by E^Cosa , _Cos8 + ^'Cos<^'Sin<^'+A:iA'' Sin8 + ^'Cos-<^'+ A" The above relations enable us to determine practically ti and 8 for any closed-circuit transformer. For on open secondary $' = 0. A' = l. D'=l+.TiCosS + ^^=l (q.p.), as it will be shown later on that Xi is always a small fraction and Ti a large number for a transformer of the type treated in this section. Hence if Cq be the primai-y current on open secondary Tl = Alternate Current Transformier. E, E, rA(l+^r:Cos8 + ^^) ''^^^ [The same is obvious otherwise, for where Lj is the inductance of the primary on open secondary]. Also on open secondary as ^' = 0, ic. tana= y=Cot8 (q.p.) Sin8+ — or Cosa=:SinS that is, //le poiver Jaclor of a closed circuit transformer on open secondary is equal to the sine of the angle of magnetic retardation of its iron for the period audflux density used. 8. In the diagram Fig. 2, we see that 0,2 is behind Cj in phase by an angle tt-/? where ;8=:0NM. Projecting ON on NM and on a line perpendicular to NM, we find that A'Cos;8=:Sin(8 + <^') + ^'Cos<^' A 'Sin^=Cos(S + the power factor of the load, the relations connecting these 8 Proceedings of the Royal Society of Victoria. quantites with E'o, Co.S(^', etc., can now be obtained as follows: — From SP Fig. 2 cut off ST so that ST = wx.jTi^crG'i = x.fi'K.f^,^ d — then ST represents —n^^ixrp-nfi.^) that is, the e.ni.f. in the secondary due to variation of its leakage flux. From IIP cut off RQ -= r^C^, then RQ represents the ohmic drop in the secondary. Subtracting the vectors ST and RQ from RS (which represents the total e.m.f. in the secondary), we get QT, which fully represents E.2, the terminal e.m.f., and the angle PQT = ^ where Cos<^ is the power factor of the load. If R be the external resistance or its equivalent in the secondary circuit so that R = Ra - ^2, and if then as d = — ^s — and t., = Ra /-2 u 1 1 1 we have _l— • n\ QP_R.,-r2_ RQ' Since Rp- " r7— i^^-^ E,^Cosc^ = E,'^'Cos<^' ^ ^'Cosc^' ;/., El or E.>= /irt r ~ -FT; (spe S 9). Again since PS = PT + TS RaCjtan <^' = R Cgtan (^ + wn^crx.f^,^ tan0' tan<^ -^=-^ + ^2 • (11.) By means of the relations I. and II. we can now transform the formulae already obtained in B' and <^' to others in B and ^. 11. Before doing so, however, it will be well to direct atten- tion to the possil)le values of Ti, Xg, ^, B\ x-^^ Xj, and SinS, as when these are considered the formulse admit of considerable simplifi- cation through dropping terms of negligible value. Alternate Current Transformer. 9 The constants tj and t^ for a transformer of 1 K.W. capacity at 50 periods would in no case be less than 1200, and it will be shown in Section III. (§ 55), that for similar transformers they are proportional to the square root of the output, and to the square root of the frequency. The greatest practical value of 6 or 6' for any transformer will not be much above V ^ where r is the mean of r, and t^ unless in case of excessive overload. The leakage coefficients Xi and x^ should each be less than .002 ; and in transformers whose coils are wound in sections and interleaved they become very much smaller. 8, the angle of magnetic retardation, will lie between 40° and 55°, its value depending on the quality of the iron, thickness of laminae, flux density and frequency ; hence SiuS will have a value between .65 and .8. [The formulse given in this paper are only roughly approximate when applied to open-circuit transformers, as will be explained further on. For them the t constants are roughly .05 -.04 times the constants of closed-circuit transformers of the same capacity while SinS=.15— .08.] 1 _ 1 tan^c^' 12. Since ^,2^^^^,^,-^,+ -^7^ 1 tan0' we find on substituting for ^, and — ^7— from equations I. and II. of § 10 that ^Cos<^ ,=M 6'Cos(/>' where M^ = 1 + 2 (x^Sincji + -^\ ^Cosc^ + ^.v/ + -,] ^^Cos ^ From § 4 we have A'!__i^2Sin(8+^) J^ 6'^CosV'~ ^'Cos<^' ' r-'CosV 2Sin8 ^ „tand)' 1 tan^' =: 1 + —^ + 2CosS-^ + ^, + -^ substituting as before we find that A"^ _ A'' e'^Cos>'~^'-^Cos> 10 Proceedings of the Royal Society of Victoria. where. A*= ^'Cos-0 ( 1 + 2a:,CosS + 2 -^ ) + ill which, for all practical purposes, I\P may be taken =1. jy'2 J)2 Similarly ^.q— y =^2^^ where, after dropping insignificant terms, D-^=l + 2:ViCos8 + 2-^ + 2^Cos(/)(XSin.^+TCos0) + 6^CosV(X^ + P) in which X = Xi + ^o We also find _ Cos8 + ^Cos0Sin(^ + XO^Cos-f *"" Sin8 + ^Cos-^./. + T^-Cos^.^ ^ , tan (h-\-OX tanA= -—. — . 1+^T 13. Transforming the equations in § 9 by means of the relations in § 12, we get P _ El A J_ F_«iCi_ «.A__Ej_ «i M o-""^ ~^Cos^~;7^ D E.,=:^ -' (see § 10). ;/i D and if P.^ be the output of the ti'ansformer P,=iEACos0 = l^^-^ also if Hj and H.^ be the copper loss in the primai-y and secondary respectively, i^l — ^^V^l — * „^2 T\2 Alternate Current Transformer. 11 where M, A, and D have the values given in the last paragraph. 14. When we multiply both sides of the first equation in § 7 by ^Cj, we get AEiCiOosa=:i/'iCi''(l + -^(Sin8 +^'CosY)\ which expresses the power P^ supplied to the transformer as the sum of three terms, of which the first is equal to the copper loss in the primary coil ; the second term *^iC;^Il^^=H,,say, A " is equal to the total iron loss in the transformer ; and the third terra /r,CosY_ is equal to the power passed down to, and developed in the secondary coil : — For, neglecting rC^ losses, the energy entering the transformer -dF on the primary side in any element of time dt is nfiy—j:dt, and the energy leaving the transformer on the secondary side in the dF same element of time dt \& — ti,f^^--j-dt, hence in the time di the transformer absorbs energy to the amount ^F //jCi + ^A-^^A of which a part ^M goes to increase the magnetic energy of the iron, while the remainder ^W is dissipated as heat by hysteresis and eddy currents. But amp. WjCi + ^aCa = F/o- SO that we may write «iCi + «.3C.2 = F - -^intvt in which case F = FSin(7t/^— 8) where 8 is the angle of magnetic lag: — F- . hence Pl=:P2 + H, + H2 + H,. It is worth noting that, as M/D = l to the first order, the iron loss H3, and the flux F will, to the same order, be constant throughout the range of operation of a transformer. 15. The efficiency rj, of the transformer being P2_ P2 -P1-P2 + H1 + H2 + H3 ^Oos> we have .; = ^~~&'O^^V~~Zr , Tj T2 ~ — ?~^' ^^^*'® O = Sin 8 + - + (9 Cos ^ j Cos ^ + Alternate Current Transformer. 13 2Sin(8 + <^) / Cos<^> + 2 {xS^n + ) Sin8 | + ^^Cos^.^ ,1 1 2 / „ SinS\ / 1 \ 1 ^1 ^2 ^1 i-l ~2 ' To find the value of 6, for which t; is a maximum when ^ is constant, we note that rf is of the form a + ljO + cO'' which is a maximum when O.^^a/c^Oa^ (say), and its maximum value is 1 Hence the value of 6 for maximum efiiciency is given by SinS+ - ^^Cos^ 0 = I^ 1 1 2/ ^ „ Sin8\ / 1\ - H |--{x,Cos8+ ) + ( AV + ^JSinS which for all practical purposes may be reduced to ., SinS - + - and the maximum efficiency is given to a sufficient approxima- tion by 7;(max) = 1 +_^ A/|i + i )sin8-f 2.T2tan(X'^ + 'P) where X.=Xi + Xi and T= -|- — Let P„= 1 ^ SinS r,r,)l+2x,Cos8 + 2 Iron loss on open secondary = sIHs ' (""' § "> Power absorbed on open secondary Power factor of transformer on open secondary ^^'"'' and take y=z:f^-^ ■^ PqCos^ so that y is proportional to the output ; also let XSin(/i + TCos^=/ XCos0-TSin0 = ^ Alternate Current Transformer. 15 and the above equation in P2 can be put into the form _ 6Cos0 •^"l+2;)^Cos<^+^^CosV (/ + ?') or j>/ = ^Cos<^[l-2/eCos.^ + (3/-^^)^^OosV] Inverting this series we get ^Cos0 =y[\+2py + (5/ + ^^)/] [where Do = 1 +/j + i(4/ + ^^)/l a very important relation, as it will enable us to transform all our fornmlae from the independent variable 0 to what is the practically important independent variable, namely, the output of the transfortner. 19. Thus if we let a= riTi( 1 + .TiCosS + j = Primary current on open secondary, the formulae in § 13 become Ci = Co V /I^o'( 1 + 2x,CosS + 2-^ ) + 2_>'Sin(8 + <^ ) + jj-^ 71 C,--iDoCoy n. F r^ */ 1 ^ i „. Cosv a small quantity of the first order, so that Ej and E^ are always approximately in opposite phases. Obviously all the above formulae will apply to non-inductive loads when cp is made zero in them. 20. The pressure drop at the secondary terminals from no load to any value of the load can now be expressed in terms of the load, its power factor, the transformer numerics and the leakage coefficients as follows : — We have (see § 19) hind- so that E, (at load given by^') 1 Ej (at no load) Dq 1 "^ Do(l+;CiCosS + ^^) l+/.j' + i(4/ + ^V = l-/r-i(2/ + ^V and the percentage drop for any load given hj y = l00j\p + hJ2p' + f)y]. Remembering the values of j> and ^ (§18) we see that the drop depends on the sum of the reciprocals of the transformer numerics (t e. on T) and on the sum of the leakage coefficients. We also see that for non-inductive loads the leakage effect on the drop is only a second-order term, while for inductive loads it is a first-order term, thus showing how important it is to have a small leakage in a transformer that has to operate inductive loads. 21. If the transformer were so designed that at full load it works with maximum efficiency, then the full load value of y or Alternate Current Transformer. 17 P^^^^(,_2^^^] (see§M5, 18). JOSd) '11 i ^ and the percentage drop from no load to full ioad in such a case is ,ooV^»{/-(/-0Vf^} which shows that the limit of possible excellence in regulation of a transformer, designed as above and perfectly wound, i.e., having no magnetic leakage, is when the percentage drop from no load to full load is lOOv/T^i^ Cos ' R = 100- jcSin(/)+ /Cos = .6 R = 2.78 which agree very closely with the observed values of R, namely 2.65 and 2.80. As in each case the full load is lOCosqi K.W. (see § 21) and the above Full load y= PoCos^ 20 Proceedings of the Royal Society of Victoria. we have at full load 10000^^, but .r=ji'X = .0218 and t=^yT = .OU ^X_ .0218 i^^T _ -016 •'• ^^ Po~ 10000 """"^ ^^ Po~ 10000 (^-^ where Pq has its usual signification (see § 18). The iron loss is given as 1-38 watts, and we may consider it as constant without introducing any appreciable error, so that PoSin8= 138.10' hence from (I.) TSin8 = --^l^-^^^ = .000221 (11.) The maximum efficiency of the transformer when Cos(^=l being very approximately = ==- (see § 15) l + 2\/TSin8 is =.971 which is the same as the observed value at full load. The copper losses Hj + Ha when ^ = 0 being H,+H,= Ip./(D,;^+?^^) •016 T>,/ri-2 138. 10^ ioxh,+h,)=joqo/4Do + -^^) where, when Cos0=l. D„^=l + 2l^P, + i(6li, + f;V./ (see§ 18) and substituting from (I.) 32^ P., _L/Z?V ^»'~-^^T0' 10'^lO"VlOV Let the load be 10000 z watts so that z is the fraction of full load then P, = 100002 10' and the copper losses in watts for any fraction z of the full load are Alternate Current Transformer. 21 Hi + H, ,,^ / 32 1 , 138v from which we deduce for the transformer in question tliat the copper losses are = 167 watts at full load = 94 watts at ^ load = 42 watts at ^ load = 10 watts at ^ load. The iron loss being 138 watts we find that the efficiency at full load is 97.04 per cent., that at f load is 97.02 „ „ ,, ,, ^ load is 96.54 ,, ,, „ ,, I load is 94.40 „ „ which figures, when compared with those in the maker's specifi- cation given above, show a very remarkable agreement. Thus we have been able to deduce with considerable accuracy from two observations of the regulation for different power factors, and the observed iron loss, the other details of the transformer given by the manufacturer. If we assume 8 = 50° which would mean that the power factor on open secondai-y was equal to SinS or .766, and take Ti = t.2 then Ti = 6930 for this 10 K.W. 60 period ti'ansformer ; eiud Xj^ + x^ = .0004. 24. As a second illustration of the agreement between the foregoing theory and practice I will consider the X'ecord of a test of a Westinghouse transformer, published in Fleming's "Alternate Current Transformer," vol. i., pp. 564, 569. From the no-load readings of Cj, Pj, and E,2 we can determine as has been explained (t5§ 7, 19, 22) tj, SinS, njfi^ and t^, while the voltage drop for any load enables us to find x^ + x^ when Ti, T2 and SinS are known (§ 22). These constants together with r-^, r,-, and the primary voltage enable us to calculate all the variable quantities connected with the transformer for any load. This has been done for the above transformer tested by Fleming, and the calculated values of Cj, C2, Pj, r/, and Cosa for each output in his test ai'e given in Table I. in parallel columns with the values experimentally obtained by him for these quantities. 22 Proceedings of the Royal Society of Victoria. As the whole behaviour of the transformer is to be evolved from the no-load readings for Ci, Pj, and Ej, and the full load reading of E.^, it is necessary that these should be obtained with accuracy. The figures for C^ and Pj given in the record of the test were probably obtained very near the zeros of the scales of the ammeter and wattmeter used, so instead of relying on single readings 1 obtained the no-load values of Cj and Pj by plotting a few of the readings for them near no-load against the output, and taking the values given by the points where the curves obtained intersect the no-load axis. In this way I find that Cj at no-load = .058 amp. and P^ = 110 watts, which values give the same power factor, .79, as Fleming obtained. It will be seen on inspection of the following table, that the agreement between the values I have calculated and those observed by Fleming is remarkably close. A very slight modifi- cation or correction of the primary wattmeter readings, which the recorded values of the power factor seem to suggest, would make the agreement almost perfect when allowance is made for the inevitable variations from mathematical accuracy of any series of observations. TABLE I. Compai'ison of the measured values of the variables obtained in a test by Fleming of a VYestinghouse transformer with values theoretically calculated by the author from no-load values and voltage drop. Power, 6500 watts. Frequency, 82.5 periods per second. ri = 5.95 ohms; ^3 = 0.0108 ohms, at 96'F. -r-, -../Nrv 1. -n riOl volts at no-load E, = 2400 volts; E,= |gg g ^^^j^^ ^^ ggg^ ^^^^^^ Ti = 6950, T,=6780, Sin8 = 0.79, -' = 23.76, [ A-i + x, = 0.003 J Alternate Current Transformer. 23 Power Factor Cosa. t ■3 0 »--oocot-ooooa)(»xoot-t~oinin-*c<5«c^.-ir-iwooo cioai^r^xxciSi c:o:3:c:^o;c:osC5 0;o:ri~c;c;sicsOic; t~ X 0 :-. =: 0-. c. S5 c ci -. =-, =: 0 0: c: o> .O! 0-. =-. c; 0-. 3: 3: oa c: 0; II r-xc: e:c:3> = = OCO = oeo 3: 0000 C-- =!=;&=: C-. 0 00000 0 --• — i-ir-..-.,-irt.-.,-HOr- — r-.,-iOC:0000.-i 1 0 ooai.-2c;q'iriir3CD:r;o«3:dc5o:ot-t-^t^i^t.^ Tj0 0 101 200 296 390 482 800 1019 1311 1802 1992 2193 2474 2966 3713 4179 4633 5000 5104 5499 5700 5867 6053 6142 0218 0317 6384 25. The method by which magnetic leakage has been dealt with in the preceding theory is not applicable to open circuit transformers. It will be seen in Section III. that this method depends on the fact that in closed circuit transformers the vector '^iC'i + w-zCa which represents the magnetizing ampere turns is small compared with either n^Ci or ti^C^, throughout the greater portion of the working range ; or, differently stated, that Cj and C.2 are practically in opposition in phase, and that WiCj — //.^C.^ is small relatively to either n-^C^ or //.^Cj from a small fraction of full load onwards, in closed-circuit transformers. 24 Proceedings of the Royal Society of Victoria. In open-circuit transformers, on account of the great reluctance of their magnetic circuits, the magnetizing ampere turns are necessarily high, and neither of the two conditions stated above are approximately fulfilled unless over a small range near full load. If we neglect leakage, or be satisfied with the rough approxi- mation to its eflfects that the present method aflFords for open- circuit transformers, then the theory given — as it is equally valid in other particulars for the two types — will apply with fair approximation to accuracy to the open-circuit type, especially in the neighbourhood of full load. There will be considerable difference, however, in the values of the constants and other characteristics of transformers of equal capacity of the two types. Let us assume that we have two transformers, one of each type, in which the cores, of the same iron, have the same cross section and volume. Let tj, t^, S, cr, 6 i efer to the closed, and Ti", T.2, 8°, (T°, 0° refer to the open circuit one. Then if they are so wound that when working under similar conditions their resultant fluxes F and F" are equal, F (a) Since - ~Siimp.{n^Ci + ^i^C^) (where o- = 47r permeance of magnetic circuit) their magnetizing ainpere turns will be inversely proportional to the permeances of their magnetic circuits. (b) Their iron losses will be equal or P2 JP02 iw— 8inS = |z£^— flSind" (see § 14) and as F = F« Sin8_8in8» o- "^^ or the sines of the angles of magnetic lag of the tzvo transformers are proportional to the pertneafices of their magnetic circuits. Again if the two cores, carrying equal fluxes, have their secondary coils such that the outputs and .secondary voltages are equal, the sections of their secondary wires and the numbers of their secondary turns will be equal, and hence the conductivities of their secondary copper belts will be equal, so that Alternate Current Transformer. 25 ''"2 Ta . .> /j\ -=^(see^6) or f/ie secondary numerics are proportional to the permeances of the ttvo magnetic circuits. If tlie closed-circuit transformer be of the core type, so that its windings are similarly arranged to those of the other, then approximately, -i^'io and hence TSinS=T"SinS''. cr a- If the method of treating leakage was equally legitimate for the two types, we have also, approximately, ^=-^ = — (see § 34), provided the windings are similar. 1 2 ^ For a non-inductive load, ^_P,_ P,Sin8 Pq Iron loss to the first order for both, hence if 6 and 6° refer to the same output, that is to the same fraction of full load in each 0 SinS a- ^u = gj^o = ^ approximately that is, the values of the co-ordinate 6 ]or the same output are proportional to the permeances of the magnetic circuits. 26. In order to compare the rates of approach to opposition of Oj and C2 in transformers of the two types, let us consider the equation Cot^ = tan(8,*).g-^ (see §19) which is correct to the first order for both, where tt - ^8 is the angle C2 is behind C^ in phase. For a non-inductive load P,^ P,Sin8 ^Pjg-^^g Pq Iron loss H3 so that Cot/3 = tanS(l +jj') for both types when <^ = 0. But 8 for the closed circuit type will be 50° or over, and for the open circuit (hedgehog) type will be about 4° ; and if we assume (which will be sufficiently accurate for our present purpose) that in transformers of equal capacity of the two types 26 Proceedings of the Royal Society of Victoria. the u'ou losses are equal, we find, taking the figures for the 10 K.W. transformer discussed in § 23, that /? will be given by the equation / 10,000 X Cot/3 = tan8n +z ,oQ ) where z is the fraction of full load and 8 = 50' for the closed- circuit, and =4° for the open-circuit transformer. The figures in the following table, calculated from the above formula, show the relative approach to opposition of C^ and C^ in 10 K.W. transformers of the two types. Fractions of Full Load - - - 0 0.1 0.25 0.5 1.0 /? for closed-circuit transformer - 40° 5° 2.5° 1.3° 0.7° (3 for open-cvirrent transformer - 86° 60° 37° 21° 11° In V)oth cases the approach to opposition will be quicker for larger transformers, as in them the iron loss is a smaller fraction of the full load output. For inductive loads having a constant power factor Cos<^ P.,Sin8 Cot^ = tan(8 + <^) + g^^^^^^^^^g^^ 7r SO that at no-load /? = ?, — (S + 0) or C^ and C, at the beginning of the range are nearer to opposition for both types than when the load is non-inductive, and as Cos<^Cos(8 + <^) is less than unity, the successive increments to Cot/3 for definite fractions of the load will be greater ; hence for both types of transformers the approacli to opposition of C^ and C, will be more marked with inductive than with non-inductive loads. Section II. 27. The theory developed in Section I. is easily applicable to the design of a closed-circuit transformer, when the full load, power factor of the load, periodicity, and e.m.f.'s are given. In the first place we would select the form of the magnetic circuit, and after consideration of the probable cooling surface, Alternate Current Transformer. 27 volume, and method of cooling to be adopted, decide on the per- missible copper and iron losses per unit volume. If K be the copper loss decided on, per second, per unit volume, at full load, then whence c, the amplitude of the full-load current density, is known, p being the specific resistance of copper at the expected working temperature. When the iron loss per second, per unit volume, (I. say) is given, the corresponding retardation 8, permeability and flux density can be got from curves similar to those in Fig. I., that have been obtained from a sample of the iron to be used with (f.p.) sine wave magnetising currents whose period was the given one. If y be the flux density ( = B the abscissae in Fig. I.), then we should have between these quantities the relation, -^ = I. (see § 1). Once tlie form of the magnetic circuit has been selected, its dimensions can be completely specified by two variables. The output at full load, P.j say, can V)e expressed in terms of tliese two variables, for P'^, the power passed down to the secondary and developed in it, is given by the equation, and F = iron .section x permissible flux density (y), fi^C'z^ total copper section x permissible current density at full load (c). P'2 can be obtained from P^ the given output, and Cos<^' from Cos<^, the given power factor of the load, by the equations, p',={i + ,^}p> , , I tan<^i and tan(^ =tan<5!)+ 1 x.^ V^, or Cos^'^Cos^- 1 - f a:.^— )Sin(/)Cos<^^j !- where 0^ is the full load value of 6, when approximate values of the transformer numeric and of the secondary leakage coefficient are known. [In future t will be used for either tj or t.j when approximate values only are required]. 28 Proceedings of tJie Royal Society of Victoria. The approximate value of r required may be found by a rough preliniiuary calculation, or from a formula such as that given in § 55, when r for some other transformer of the same type is known, and 0^, the fall load value of + fi), and the mean length of a turn of either primary or secondary coil (so wound as to be the same for botli) is 8 (d + fS). If /J the space factor of the iron be taken =.9, then the cross section of the iron circuit is =4/^'^ = 3. 6/3', and the volume of the iron = 1 6//3-( 2(^ + ^) = 1 4. 4/3'^( 2^ + /?). The space factor / will not only enable us to allow for insula- tion between the laminae, but also for ventilating or cooling ducts, if such are deemed necessary. Let us decide that the iron loss shall be 10* ergs, per second, per unit volume. With a sample of transformer iron .045 cm. thick I have found with my wave tracer,|| when the iron loss per cm.^ per second was 10'' ergs, at 50 periods, the magnetising current wave form being slightly peaked, that y = 4800, 8 = 52° iJL = 2250 (q.p.) so for the present design I will assume, for the iron to be used, * Called theoretical because the details are full)' worked out iti accordance with the theory already given. A knowledge of the theory and experience will, however, enable one to make sufficiently accurate allowance for most of the small correcting terms, instead of having to calculate them in each particular case. f This is far from being the most efficient shape, as will be shown in Section III. t The different factors and constants assumed are not given with any authority. The purpose of this part of the paper is merely illustrative. II Phil. Mag., Nov., 1903. 30 Proceedings of the Royal Society of Victoria. that y = 4847, 8=50° yii=2250, which satisfy the sine wave equation Zf;y-Sin8_^^^ OTTfX which states that the iron loss per cm.''' per second is 10" ergs. The core loss will therefore be 144.10^/3-(2<^ + /3) at full load. 29. The kind of winding to be adopted will depend on the excellence of regulation for inductive loads that is required. In Section III. will be shown how to calculate the leakage coefficients for different kinds of windings, and how the regu- lation for inductive loads that these windings will give, may with considerable accuracy be predetermined. A very important consideration with regard to the arrange- ment of the windings is clearly shown by the general expression for the efficiency given in § 15. It will be seen that in the denominator, positive terms depending on x,, the leakage co- efficient of the secondary, occur, of the first order in small quantities for inductive loads, and of the second order for non- inductive loads. Hence it is obvious that if we have a choice, we should so place the secondary windings or sections that x^ is the least possible. Now it will be shown in Section III. that, with interleaved windings symmetrically arranged with regard to the middle line across the window, which line must there- fore bisect the central section of one of the coils, the leakage coefficient of that coil to which the two outer sections belong is negative. Hence the most efficient arrangement is that in which the two outer sections belong to the secondary or output coil. The regulation will be the same whether one or other of the coils has the outer positions, as it depends on .Tj + Xj, which is little or not at all affected by the interchange. Assuming that for the present design a winding in three sections will give satisfactory regulation, we will place the whole primary coil as a single section in the central position between the two halves of the secondary coil. For such a winding, when the iron and copper losses per cm^, and the space factors are as we assume in this design, and when /x = 2250 (see § 51), Alternate Current Transformer. 31 ,ri = .00129 .T,= -.00024 .Ti + .r2=. 00105. 30. If Kj and K.^ be the permissible copper losses per second per cm.'' at full load, Ki for the primary being decided on as say 15.10^ ergs, and Kg differing from Kj by a small amount which will depend on how the copper losses are to be divided between the two coils ; then where c\ and c^ are the amplitudes of the current densities, and p the specific resistance of copper at the expected working- temperature. Let us take /5=1800 abs. then fi = 12.91 abs. Let j-j, ^2) ^^e the sectional areas, /j, 4 the mean turns, and ^1, ^2 tlie space factors of the two coils ; so that ^iXj, j^.^^a, are their total copper sections, and ^iS-^/i, ^^^.J^i their total copper volumes; then (.see § 6) since /j^/i, = ^1 + K say, ^2h T2 where k is a small quantity to be determined, depending on the current densities in the two coils at full load. This equation, together with Si + s.^ — i^'^, give us the copper sections, ^252=2Q/;^{i-^k} where Q= — -— ^the harmonic mean of a, and Oo. Let us assume for the copper space factors ^i=.5, ^2=.7 then Q=.583 and ^i5i=l. 166(1 +.42k)^^ ^^j^= 1.1 66(1 -.58k)^^ 31. If we arrange so that the current densities in the two coils shall be equal at full load, then 32 Proceedings of the Royal Society of Victoria. and ^i^!^=.l+^^"(^ + ^) (see §13) q^s^ fi^C.2 6,Cos Sin(8 + ^) where C^, Co, and 6^ are full load values. In addition let us arrange that the copper and iron losses shall be equal at full load. Then (see § 17), as 2=1, SO that K=Sin(8 + 0)^ 2 For the determination of k and other small correcting terms, an approximate value of t must be known. We can easily obtain one by a rough preliminary calculation in which these correcting terms are neglected, or from the formula given in § 55, when r for some other transformer of the same type is known. The first method gives us t=6000 ; hence as S-=50°, Sin8-=.766, Cos(^ = .8, (/)=37', we find that, ^,Cos0=48, ^,=60, K-.02, T,/T,= l.02, ^^=.01 T and taking ^3= -.00024 (see § 28), Cos(p'=.808 at full load (see §§ 10, 27). Substituting the value of k thus determined in the expressions for q^Si, and q.^s., we get ^15-1^1.176^^ q.2h=^-^^^^^- Hence, the total volume of copper being =/(^i^i + q,s,)=8{A + (3){q,s, + q,s.,), it is = 18.63//^(/' + /3), and as the copper losses at full load are = 15.10"' X volume, Alternate Current Transformer. 33 they are =2795.10=^^^(^+^). 32. As we intend that the copper and iron losses shall be equal at full load, 2795^-(^ + g)^^ 3^ or J^ .5152 ?r(« + 2) where u^/3ji>, from which equation ?e (the positive root) can easily be determined by trial (using a slide rule) and is =1.151, hence y8=1.151^. 33. The output P^, being 10 K.W., that is, 10" ergs, per second, and (see § 13) as we find, using the approximate value for — given in § 31, that P'.j, the total power developed in the secondary is = 1.01 . 10", but P'.,=iE'2C,Cos0' =hzvfi.:^C.2Cos(f>' ; hence, using the value of Cos(/>' given in i^ 31, and remembering that w=2-nr . 50=1007r, 202 . 10'-' "■2^2^=100^ . .808' =796 . 10". Now C2 being the permissible current density = 12.91 (see § 30) «A=^2^2-f.= 12.91 . 1.1 53<^-^ = 14.9^^ and y being the permissible flux density=4847, F=4;^/5'-^y=3.6 . 4847 . ^\ = 17450/3'. u- ,.,^, 796 . 10' «""" ^"^ -14.9 . 17450-="^«^ and as ^==1-151^ we find that /3=7.98 and ^=6.93, which determine the carcass of the transformer. 34 Proceedings of the Royal Society of Victoria. Substituting these values of (3 and ^ in the expi-essions for the copper and iron losses at full load given in §§ 31, 28, we find that each is equal to 200.3 . 10', that is to 200.3 watts. Hence the efficiency at full (inductive) load of 10 K.W. will be 10000 ^, ,, 10400.6^^ percent. 34. The numerics tj and t^ can now be determined for as T-^=z4:Trzv X Conductance of primary copper belt x Permeance of magnetic circuit "■i— *''^8(^ + ^)p 4(2^ + /3) 400 . Tf" . 1.176 . 48.08 . 3.6 . 63.7 . 2250 ~ 8 . 14.91 . 1800 . 4 . 21.84 = 6140, and as Ti^1.02t2, T,=6020. From the results already obtained the accurate full load value of 6JJos(f> can now be calculated by means of the formula in § 17, remembering that z=\. Thus we get at full load 6,Cos^ 13, E,= «2 E, or n and as D= 1.044 (see §35), «.2 = :89.16. 37. The sectional area of the conductors to be used being ^'^and£^^ for the primary and secondary respectively, we find (see § § 31, 33), Sect, area of primary conductor=0.0662^;;/^. Sect, area of secondary conductor=0.6218f///^. The mean length of a turn of either coil being =8(/^ + /5) = 119.3 and as /3=1800, the resistances of the coils (warm) are Primary, 2.77 ohms. Secondary, 0.0308 ohms. 38. The terminal voltage at no load being (see § 13) fh El «i , ^ ^ SinS ' 1+a:iCos8+ is =229.5 volts, and as it is 220 volts at full inductive load the drop will be 4.14 per cent. If the same transformer operated a non-inductive load it would be rated as of — -= 12.5 K.W. capacity, and its voltage drop from no load to a non-inductive load of 12.5 K.W. would be 1.5 per cent. 39. By means of the formulae obtained in Section I. the curves shown in Fis. 3 and Fie;. 4 were constructed for the 36 Proceedings of the Royal Society of Victoria. transfotmer we have designed. Those in Fig. 3 refer to it when operating the kind of load for which it was designed, namely, one with a constant power factor of .8, while those in Fig. 4 refer to the same transformer when operating a non-inductive load. A comparison of tliese theoretical curves with similar ones obtained practically from actual transformers will afford a further illustra- tion of the agreement between the theory I have given and practice. 40. In the preceding design it was arranged that the current densities in the two coils should be equal at full load. Any other desired distribution of current density, however, can be equally well dealt with. Thus if Kj and K.^, the copper losses per c/n'^ at full load, be each given, we know d and C2 as K=^pf^, and K is determined as in ^ 30, by aid of an approximate value of t, so that the ratio ?-^ can be found ; and as i'i + i'.2=4(6'" we can determine Si and s., and proceed as before. Again if K^ be given, and we wish to arrange so that at full load the primary and secondary copper losses shall be equal, we have ^UlI^I+k (as above), and ^i5'i/iKi=^2-S"24K2 or (/iSiCi^=^2W as /i=4 *i^cl Is^^^pr^ hence ^^={1+k)% which with s-^ + s^^Aib' determine j'j and So, and we proceed as before. In transformers of the core or H type, in which the primary coil almost completely surrounds the secondary, /^ is greater than 4, and the preceding method would have to be modified. Alternate Current Transformer. 37 When Si and x^ have been determined as before, /j and /« can be expressed in terms of the two variables d and f3, and the ratio of the losses I (volume of iron) being equated to the selected value s gives us an equation, slightly more complex than that for a shell transformer, for determining (3/b and the rest follows as in the preceding cases. 41. When we select for the section of the iron tongue and for the windows or winding apertures different shapes from that selected in ^ 28, the method of procedure is fairly obvious. In general if 2^, 2/3', be the section of the tongue, 2y8' being measured perpendicular to the planes of the laminae ; and if 2d, 2b' , be the dimensions of the window 2b' being measured parallel to the tongue; the volume of iron =16//3/3'(/^ + '^' + /3) and the volume of copper ^16Q/^^'(/3 + y8' + 2^) [neglecting the small correcting terms in k depending on the dis- tribution of current densities in the two coils], whei-e / and Q have the same signification as before, and if the iron and copper los.ses are to be equal at full load bb\^ + ^' + 2b)_ p\_ .9.10-' _i Q.,9 pli'{b + b' + li) QK .583 . 15 . 10^ if we adopt the same values for the data as before. The values of [ijb for a few special shapes are as follows : — [a) lib=b', 2/3'=3/3, /3//^=.984. (/;) If b=b\ (S'=2f3, f3/b=.886. (c) If 2^'=3/;, 2/3'=3/3, /3lb=l.Ul. {d) If 2b'=?>b, /3'=2/3, ^/^=1.025. {e) lib'=2b, 2/3'=5^, /3/^=1.042. Let us determine approximate values of t for transformers of the above shapes whose output on non-inductive load shall be 12.5 K.W. at 50 periods, the normal rating of the transformer already designed. 38 Proceedings of the Royal Society of Victoria. We have (see ^33) ]f2vn.2G^=l'2.b . 10'° + Secondary copper loss, and the secondary copper loss may in this connection be neglected in a rough deter- mination of T ; but «A=^2 . ^ . Ub'=\b.QUly\ F=y . p . 4/3/3'=17450^/?', taking the values ^2=12.91, y=4847 already used; hence 12 5 1 0^° bb'BB'= • =3030, ^^ 507r . 15.05 . 17450 which with the ratios ^jb above enables us to determine (i and b in each case. The formula for the numeric t can be put in the form _ TT/x Output _ 3,925.000 ''~2pcy {/3+(3' + 2b){b±b' + /3)~{^+p' + 2b) {b + b''+ P) by means of which its values in the five special cases considered can be determined. Thus we obtain the following details given in tabular form. (a) (?>) (c) (d) (e) b 6.76 6.63 5.67 5.57 4.86 b' 6.76 6.63 8.50 8.35 9.72 (3 6.65 5.87 6.47 5.71 5.06 /?' 9.97 11.74 9.70 11.42 12.65 Volume of copper Volume of iron U2840 i 19250 12660 18990 12450 18680 12280 18420 12090 18140 T 6460 6650 6910 7080 7280 We also find that the iron losses at full load, which are half the total losses, are for (a) 192.5, (/;) 189.9, (c) 186.8, (d) 184.2, and (e) 181.4 watts, so that transformer (e) is the most efficient of the series, having an efficiency at full load of 97.2 per cent. Obviously the transformer designed in detail with square windows and square tongue is of a less efficient shape than any of these, as its iron loss at full load is 200.3 watts. The volume of iron in each of the present series in cub. cms. is 100 times the iron loss in watts. In {e) it is 18140 cub. cms., Alternate Current Transformer. 39 which corresponds to 25 lbs. of iron core for each kilowatt of full load activity. It is worth noting that, for transformers of the same type made of similar iron, the percentage iron and copper losses and the weights of copper and iron per kilowatt of full load activity, are inversely proportional to the fourth root of the product of the output into the frequency. Section III. Transformer Leakage. 42. In addition to the magnetic lines forming the main flux F, produced by the combined action of G\ and C.^ and looped on both circuits of a transformer, there are other lines, the leakage lines, that are only partially looped on the two coils and that traverse the space occupied by the coils, in some cases completing their circuits in the iron. It will be shown that, after the transformer is somewhat loaded, the effect of these leaknge lines on its opera- tion is the same as would be produced l)y two fluxes ; one, the primary leakage flux, supposed to consist of lines in phase with the primary current that are looped on the primary and miss the secondary circuit, and the other, the secondary leakage flux, suppposed to consist of lines in phase with the secondary cui'rent, that ai-e looped on the secondary and miss the primary circuit. 43. Let Lj and L.j be the inductances of the two coils when the leakage lines only are considered, Mj.^ the mutual inductance of the primary on the secondary, that is the number of leakage lines looped on the secondary arising from unit current in the primary, and M21 the mutual inductance of the .secondary on the primary. (Mj^ will not in general be equal to Mji). The e.m.f. e^ generated in the primary coil by variation of the leakage lines due to C^ and Cj will be represented by the vector zt/LiCj + K'M^iC.j after it has been turned through a right angle in the negative direction : but since the vector which represents the magnetising ampere turns, is, after the 40 Proceedings of the Royal Society of Victoria. transformer is somewhat loaded, negligible in comparison with either //iCj or ti.f^., (see §§ 9, 13, 25, 26), and hence e^ is the vector w I after it has been turned back through a right angle and amp. e-i^ = 2v\ ~— — fi V«jCj Thus we see that the eifect of the leakage lines on the primary circuit is the same, when the transformer carries a load, as would be produced by a flux looped on it but not on the secondary ; but in § 2 this flux was specified by XiO-n-^C-^^ where x^ is the primary leakage coefiicient, and o- is 47r times the permeance of the magnetic circuit, hence Similarly if x^ be the secondary leakage coefficient 44. Let us determine Lj, L2, M12, M21 and thence jv, and x^^ for a shell transformer in which both primary and secondary coils are single. This must be done in two parts. We must determine the values of those portions of the above coefficients that are due to the leakage lines that cross the windows, as well as the values of their remaining portions that are due to the leakage lines that cross those parts of the coils that are not embedded in the iron. Let L'l, L'2, M'i2, M'21, x\, x\ be the former, and L"i, L"2, etc., the latter portions of the above coefficients. Let 2/3 be the width of the iron tongue measured in the plane of one of the laminae from window to window, and let 2/3' be its height measured perpendicular to the laminae. Also (see Fig. 5) let PP' the breadth of the windowz=D, PO the thickness Alternate Current Transformer. 41 of the primary coil=/'i, and SO the thickness of the secondary coil=A2. In order to determine L\, L'2, etc., consider first the leakage lines produced by C^ that cross the windows through the spaces occupied by the primary coil. If we draw two planes A' A, B'B, Fig. 5, the same distance s on either side of the median plane of the primary, the magnetic forces due to the two portions of the primary coil AP and BO neutralize each other within the space AA'BB', so that the M.M.F. round the circuit AABB' is due to the portion of the primary it encloses and is and the flux circulating in this circuit through the elemental rectangle 2p'dz at z is as it crosses the window twice. This flux encircles the current and as the energy associated with naturally looped flux an(i current is ^ flux x current, the energy of Cj due to the lines through the space occupied by the primary coil is •I -6 D^<^^-^ ^'-^ Thus we see tliat the energy of Cj due to the leakage lines that cross and recross the space occupied by the coil carrying Cj is equal to ^ Jh^C^^ x permeance of this space across the window. Again the M.M.F. due to Cj, in any circuit that crosses the space occupied by the secondary coil, and completes itself through the iron round the primary coil is uniform, and =47r«iCi and sends through the secondary space (section 2/3'^^) the flux 42 Proceedings of the Royal Society of Victoria. (neglecting the reluctance of the iron). This flux encircles all of n-^Ci and the energy associated with it is SO that the total energy of Cj due to the leakage lines it produces and that cross the window is The other window contributes an equal amount, hence as the sum is also equal to i-L'iC'^, L': .i-Mr.^io. ^,+ 121^ Similarly, • (II.) The latter of the two fluxes already considered is partially looped on w^Cj and the inutual energy of Cj and C.^ due to this can easily be found as follows. The flux through 2(3'dy where j=OE, Fig. 5, is and it is looped on the portion — r- y or ;/.,U., so that the mutual energy is Integrating from y=o to y^=b.2 we find that the mutual energy of C2 and the leakage lines due to C^ is Thus we see that the mutual energy of the uniform Cj flux that goes through the C^ coil, and fi.^j^^i is =\ . flux . «A- (III-) or is the same as if all the flux through the space were looped on half the total current in the space, a result that will be made use of in § 45, b. klternate Current Transformer. 43 Hence, as the other window contributes an equal amount of energy Q of \ M'i,Ci C,= -g^«i«,OiGA M ^.y= —rr'fiifi-Ai- Similarly, D 87ry8' y (IV.) M'3i= ~jy~n,;i.,/>i. Substituting from (II.) and (IV.) in tlie equations for A-'^^and ., in ^43, we get ^ ^^QAI and IGtt/?' crD r 5 1 / (V.) ;<«8,/-* -^•^ •\^i XiH-.r.,=: 287r^' 3o-D {/^i + /^. 45. The determination of the coefficients x'\ and a:'., due to leakage lines other than those that cross the windows can only be approximate. A fair approximation can, however, be obtained by assuming that these lines form circuits like aa'cb'ba, Fig. 6, of which ba is in the iron, as the inner surfaces of the coils bear against the iron tongue, aa and bb' are parallel to the plane of separation of the coils and acb^ is a semicircle in the air joining a and //. Let us consider the lines due to C^. If aa and bb' are a distance z on either side of the median plane of the primary coil, then the M.M.F. round the circuit aacb'ba is which will send through the magnetic circuit at 5, whose breadth is dz the flux 47r«iCi B^3 if B be the mean width perpendicular to dz of this elementaiy circuit. Now where the coil bears against the iron the width is 44 Proceedings of the Royal Society of Victoria. 2/3, and at the outer surface of the coil the width is 2/3 + 7rD, if we assume that the corners of the coil are quadrants of circles. Hence B = 2yS + ^D. [This allowance for the corners of the coil may be considered rather large, but if so it is compensated by the fringing of the lines crossing the windows, which has not been allowed for.] Adding an equal amount for the other uncovered side of the coil, the flux across dz due to C^ is 16B ^ zdz ■X T2D' IT and it is looped on ^^22 of ;/iCi, so that the energy associated with it is Integrating this between the limits z = l\l'2 and s^O we get 4B«,^Ci-^|i - 2A, + 4Vlog^|^|, where A,=^^ which is the energy of Cj due to its lines through the primary coil. Again the energy of Cj due to the lines it sends through the space occupied by the secondary coil is V^" =4B«/Ci'^ f J K dy 2D J -^ TT = 4B«/Cinogl:tl^^^:^i ' ' * l + 2Ai where j/j^— . K The sum of these two results being \i\l\Q,{- Similarly, if ^2=^, i/.^= i=_ TTOn p., J/, Alternate Current Transformer. 45 L „^bB»J-{ I — 2a„ + -tA^ log ' "" + 10'' -'- — — - In a like manner we find that M",,=8B.;,;/,|l - (A,+ iv,)lool + ?^±^| and substituting in the equations for x\ and x.^ in § 43 " SB/ 1 .^, ,,..,, 1 + 2A, , , l + 2Ai + 2vi , '■^"1=— -^-2Ai + 4A-ilog J +log- , ,'T + o- I JAi i + 2Ai (^i + Avi)log— j_j_2^;^ / on r > " ^'''2= \ Interchange Aj and A.,, I'l and v., in above. - and finally for the leakage coefficients of the transformer lA-i — oc 1 — r*^ 1 46. From the expressions in §^ 44, 45, let us determine x-^ and Xo, for transformers with square windows and iron tongue of square cross section, and of which the secondary coil occupies three-fourths as much space as is occupied by the primary. Then D=<5i+/;,, /^,=|d, ^.,=^D, y8=/3', v,='ili, v,=4/3, Ai=^=l.ll, A.,= ^=1.48 TT^l 77(72 From § 44 we get ^^=9.6^, a:',= 19.8^, for a shell transformer with interleaved windings we have, as before, to determine x\ and x'.j^ due to the lines that cross the windows, and x'\, x''^, due to the remaining leakage lines, separately. We will first determine x'l and a'j in the general case. Let the sections of the two coils be arranged, as in Fig. 7, symmetrically, with regard to the median plane across the window so that a section of one of the coils occupies the central position in the window. Let the coil to which the central section belongs be called the even coil and its sections the even sections, and let the other coil be called the odd coil and its sections the odd sections. The even coil will have an odd number, / say, of sections, and the odd coil will have an even number, J say, of sections where J may be equal to either / — 1 as in Fig. 7, or to /+1, in which case the two outer sections would be odd ones. Let d.^ be the thickness of each of the even sections and l>j that of each of the odd ones ; and let L.^, x^, n.^ C.2 refer to the even coil and Lj, x^, fi^, Cj, etc., to the odd one. Let 2/3, 2j3', be the section of the iron tongue, 2/3' being measured perpendicular to the laminae ; and let D be the breadth of the window. In Fig. 7 is a diagram of the window in which the sections are numbered in accordance with the above plan, the central section being numbered O, and directly below is what we may call the M.M.F. diagram for C, in which the zig-zag line AUeimate Current TransforTrier. 47 OP gives the distribution of M.M.F. in the different sections due to C.^. Thus if the M.M.F. round the Central section O, that is iL_5, be called 7/1.2, I the M.M.F. round the magnetic circuit 1,1', is fn^, the M.M.F. round the magnetic circuit .3, 3' is 3;/^,, and so on, so that the ordinate of OP opposite any odd section is proportional to the number of the section. Ajid it is easily seen that the mean ordinate of OP opposite any even section is also proportional to the number specifying that section. The flux per unit length across the window at any point in OQis =M.M.F.-^' 2D and it fn-iY) ^■^- • — 7)^^/2 it is easy to see that, due to C.,, the total flux round the circuit 1, 1', is fj?-y 2 2' is ^fh 55 55 55 55 55 55 -"5 -^ 5 '* -'Jiyi 55 55 5) 55 55 55 "^5 "^ 5 ^^ ^JP\ and so on. iyd) The energy of C2 due to the leakage lines it sends through the even (its own) sections is, as we neglect the reluctance of the iron, the same as if the odd sections were removed and the even ones pushed up together. By § 44, (I.), this energy is = g ~=Q «2^C/. (I.) The energy of C.^, due to the leakage lines it sends through the odd sections, is, by aid of Fig. 7, found to be =^D" -p~ ^^ 6 (^^•) This follows easily when it is noticed. Fig. 7, that the flux ^/^d^ (for instance) round 3, 3' is looped on three even sections, i.e., on the current Sn^C^/i, and similarly for the others. 48 Proceedings of the Royal Society of Victoria. Adding I. and II. we get the energy of Co, due to the leakage lines it sends across one window. The other window contributes an equal amount, hence, or L',= -3^ »^\t^+-^^^., ^^,j. (III.) «iC, r , 2) (I?) The mutual energy of C.^ and Cj, due to the C.j lines through the even spaces being looped on Cj, can be read off from Fig. 7, and is J J^ n,n.j:^,G'^b,. (IV.) The mutual energy of C.^ and Ci, due to the C.^ lines through the odd spaces being looped on Cj is = j^ ;/i;/,C,c/-^ b,. (V.) [This will be seen by considering circuit 3, 3'. The flux round it is "ifipi, and it completely encircles the two odd C, sections 1 and 1', and partially the Cj windings in 3 and 3'. But in § 44, III., it was shown that the mutual energy due to a uniform flux encircling a uniformly distributed current occupying the same space, was the same as if the whole flux encircled half the current ; hence, in this case, the energy contributed by 3 and 3' is the same as if the flux 'i/.^b^ encircled one only of them com- pletely, so that altogether the flux ?>fj'i encircles ?>n-^C-^lj oi the current Cj.] Adding IV. and V., and doubling the sum to allow for the other window, or M',,=M.,., \^tzlb,+ -^Izi/., j (VI.) (c) In a similar manner, by aid of the ]\I.M.F. diagram for Cj also given in Fig. 7, or from symmetry we And that Alternate Current Transformer. 49* VII. (d) Substituting the values for L',^, L'i,M'i.2, M'.i, in V., VI., and VII. in the expressions for x'^ and x\ in § 43 we find I. If the two extreme sections in the window are even ones, in which case j^i— 1, that "•^=-3^)1 ^•^+7-^^)' , , _87r/3' H where H =/7;o + (/— 1 )b^ = total height of window. II. If the two extreme sections in the window are odd ones, in which case /=/+l, ^'^-3o^i-+ r'y 47r/37/-l \ ^^"i=-3^l-+l'^^+M (c'). From the results in (d) we see I. That the coefficient x' <^ or x\ of the coil to which the two extreme sections belong is negative, indicating that this part of the leakage produces on the coil, with the extreme sections, a capacity and not an inductive effect. The same thing is true with regard to the coefficients x'o, and x\ due to the leakage lines that do not cross the windows and which will be determined in the next paragraph, and it has been pointed out in § 29 that secondary leakage reduces the efficiency of tlie transformer when operating inductive loads, hence the winding of a shell transformer for inductive work should be so arranged that the two extreme sections belong to the secondary or output cod. 50 Proceedings of the Royal Society of Victoria. II. That the sum x\-\-x'.2. of the coefficients due to the leakage lines crossing the windows is inversely proportional to the product of the numbers of sections in the two coils. (/). If 1l)=D be the breadth, and 2b' the height of the window, 2/3, 2^' the section of the iron tongue as before, ^i and q.2 the space factors of the coils; and /a the permeability of the iron ; the formulae in {d) can be put in the following forms suitable for calculation. ^'C^ + ^' + iS) q, + q, ^' Ji, 1- ±1 where the upper signs are to be taken when the two extreme sections belong to coil 2 (the even coil), of / sections, to which the middle section belongs, and the lower signs, when the extreme sections belong to coil 1. Evidently /=F 1 is the number of sections of coil 1 . 48. When we make the same assumption as is made in §45 with regard to the paths taken by the leakage lines that do not cross the windows, we can obtain the values of the coefficients x"i, x'.^ due to these lines in the general case of interleaved windings. Specifying the coils, sections, and dimensions of the trans- former exactly as in the preceding paragraph, and letting B = 2l3+'^D=2fS+7rb (see §47/). _2I) _2D we find, after proceeding as in §§ 45, 47, that 32Br 11 1 Aa X./, 1 + 2X., n=i—l ^U ^7 ) + 2tj j ^'^"2A,+«(l+v,)-l Alternate Current Transfor^ner. 51 m=j-l =1 „_32Br_l_vi_A, ^"^1- o- L 8 8 2y^ ^l 2z>- "^4/j ^°^2A,+«(1+V2)-1 j«=j— 1 ^ ( /;/(! wvi + XQ , / hmv, + Xi ^ '^-^ ^ 2X^ + w(l+Vi) + l-| + ^i "2/>- +(" y )3^°g2A, + ;.(l+.0-lJ- 7n = l where, in each of the above expressions, the values to be given to n are all the even numbers from 2 to I'—l, and to m all the odd numbers from 1 toy — 1 inclusive. It will be easily seen, by considering the case of a winding with three sections, that by spreading out the free ends of the sections as is done for cooling purposes the values of the coeffi- cients x'l and x".^ will be slightly increased. 49. If the i-esults in § 48 be written o- " ■i:7riJij3(S' ' a- ' iirn^^' ' ' it is easily seen that Xj and X.^ depend only on the shape of the window, the numbers of sections of the two coils, and the ratio of their space-factors. In the following table are given the values of X.2 and Xj for some different values of / and J, for square windows and for oblong ones whose height, measured parallel to the iron tongue, is twice their breadth {2/>'^=4-b), and for some different ratios of space-factors. It will be noticed in the following table that tlie coefficient of the coil to which the extreme sections belong is negative, and that the sum of Xj and X.^ is (q.p.) inversely proportional to the product of the numbers of sections: also that the change of the space-factor ratio from 4/3 to 3/4 does not cause much change in X] and X.2, unless in the three-section winding. Hence, by aid of this table we can obtain very approximate values of 4 a 52 Proceedings of the Royal Society of Victoria. Xj and X.J for otliei' windings, and for windows of other shapes by inteipolation. Numbers of £ Square Window Oblong- Window Sections. .2 If — 3 6 = 6' 6' = 26 Coil 2 Coil 1 xi ! X, X1+X2 Xi Xi Xi+X> ' iTl ^ 1 1 2 3/4 —.38 1.75 1.42 —.41 2.75 2.34 1 2 4/3 —.37 1.92 1.55 —.58 3.03 2.45 a 2 4/3 .76 —.29 •47 1.13 —.41 .72 3 2 3/4 .80 —.32 .48 1.21 —.48 .73 3 4 3/4 —.27 .50 .23 —.38 .75 .37 3 4 4/3 —.29 .53 .24 —.41 .79 .38 5 4 4/3 .36 —.21 .15 .54 —.32 .22 5 4 4/5 .38 —.23 .15 .56 —.35 .21 5 6 6/5 , ! —.19 .27 .08 —.29 .44 .15 [Note that tlie central section always belongs to coil 2 of i sections, i odd.] 50. There is also magnetic leakage due to the lines in the copper conductors, and in the spaces between them, which is of considerable importance in the case of large, low-pressure trans- formers. It is well known that the inductance per unit length of a wire, due to the lines in itself is 1/2, and due to the lines between its surface of radius r and a concentric cylinder of radius r' is 21og^-';V. In the case of insulated wire wound in a coil it will be very near the truth to take for r' the radius of the circle equal in area to the total area allowed each wire in the winding, so that if g be the space-factor and the inductance of the wire per unit length will be i + log-- Hence, if / be the mean length of a turn of either coil, their inductances arising from this cause are L"V=«,/||+log^^^} Alternate Current Transformer. 53 L"',=«,/|i + log-|, and if x"\y x"'.2 be the corresponding leakage coefficients •^ 1=-^ = — U + log-f. L"., / (, 1) Now it is easy to show by means of i-elations already given, that where P.2=fu]l load (non-inductive) output, t-=:amp. of current density, p=sp. res. of copper ; so that we have ze'Po ( 1 ") It will ])e shown, for similar transformers designed on the same lines, that t is proportional to v^ZfPa, hence if the e.m.f.s remain fixed, x\ and x" .^ will increase as the square root of the product of output and frequency increases. If tlie conductors be rectangular in section instead of circular, the above expressions for x-l" and A",/" will be sufficiently accurate for all practical purposes. Note. — The connectors from the ends of either coil to the corresponding ter-minals outside the cases of large transformers ought to include as sninll an area as possible, since on account of the proximity of the ii'on of the transformer and of the case, the loops so formed would have considerable inductance, thus increas ing the leakage coefficients, especially that of the low pressure coil, and so impairing the regulation on inductive loads. 51. Let us determine the leakage coefficients xp, Xg^ for the transformer designed in Section II. [In this paragraph Xp^ x g^ will be the primary and secondary coefficients respectively, while X2, q.21 etc., will still refer to the coil with the middle section]. The details for this transformer are (see §§ 28 et seq.), /=!, b=b\ /5=/3', y3=1.151^, /x=2250, p=1800, ^=12.9, Ei=3111.10^ E,=311.10«. 54 Proceediiu/s of the Royal Society of Victoria. P2=:12.5.K.W., T=6080, «'=1007r, and as the middle coil is the primary one q.j;=Lq^^=z.f)^ ^1=^*- =.7. From § 47, f, taking the lower signs, x\=^-^^ xi±] =.000490 3/.^ 2 3/./? 2 From § 49, taking from the table the values of X.2 and Xj, / = 1, b=:b{, and ^.,/^i = 3/4, which is sufficiently close to 5/7, we 47rya^^ .",= _(M±-M2^±A).33= -.00015 47r/*;8^ From § 50, ^"',=y' V = ^;( 1 + 21og2)=.000002, ."-.'".= ^(l4-21og^) = 000015, which in this case of few sections are relatively negligible. But Xp^=x'.2-\-x".2.-\-x"'.2, x^^=x\-\-x'\-\-x"\, hence Xp =.00129, x., = -.00024, Xp +Xs =.00105. If this transformer had been wound so that the secondary as a shigle coil occupied the central position, with half of the primary ozi either side, its leakage coefficients x'p^ x'^, would be y^= -.00029, .r', =.0014. and it is interesting to find what effect this change in the relative positions of the two coils would have on the efficiency for inductive loads. When we neglect all small terms but the one that depends on the first power of the leakage, we find from §15 that the maximum efficiency 1 1 + p — "7 A/TSin8 + 2a:s tan<^Sin8 = 1 1 -Q^ VTSmS i(l - -Ixs tan^SinS) := 7;o(l — 2aj,. tan^SinS) il-V-) Alternate Current Transformer. 55 where T=:|— + — =^- Tl To T Hence 7] is proportional to 1 — 2xs tan^SinS and the ratio rj/rj' of the efficiencies of the same transformer, but differently wound as above, is for loads of the same power factor, r) 1 — '2xg tani^SinS rj' 1 — '2x'g tan^SinS =zl — 2{xs — x'g )tan(^Sin8, = l + .OO328tan0Sin8, as X, =—.00024, x's =.0014. If Cos0, the power factor of the load, be .8, tan0:=.75, and "^ =1.0019 (taking 8=50°), V and if Cos(^=.6, tan^ = 4/3, and ^=1.0034. V So that when the secondary coil occupies the two outside positions the transformer will have for inductive loads, when Cos0=.8, a greater efficiency by .19 per cent., and when Cos0=:.6 a greater efficiency by .34 per cent, than when the primary occupies the outside positions. For non-inductive loads the difference in efficiency will be very small, as it then depends on the square of Xg, 52. In § 41 it has been shown how, from the data for any particular design, the dimensions of the carcass and approximate values of the numeric t and of the efficiency can be quickly obtained. Selecting from the series in § 41, transformer (c), of which the the details are : — - Capacity 12.5 K.W. at 50 periods, 2^'=3^, 2/3'=3/3, f3=lMU, t=6910, /*=2250, ^=12.9, Sin8=.766, ^^=.5, ^,=.7, to which we will add E^=2200 volts=3111 . 10^ E, =311 . 10«. let us determine its leakage coefficients and appi'oximate values of its voltage drop for different kinds of loads if it be wound in five sections, three secondary and two primary. 56 Proceedings of the Royal Society of Victoria. As the middle and end sections belong to the secondary coil, ^2=.7, ^i=.5 and from the formulae in § 47, /., taking the upper sign, we get .t'.,= -'— ^=-.000168, .909 x\= = .000404. From the formulae in § 49, for X^ and Xj, which are to be for a window in which ///^=:1.5, we will take the mean of the values given in the table for win- dows in which d' jb^=\ and b' jb^^l, and as /=:3, five sections, W?i = 4-/3, we get .29 + . 41 X.,^-"^- =-.35 .76 + 1.13 Xi= — ^ =: .945 hence a;"2=-.000125, a;"i=. 000338. From the formulae in § 50, a;'%r=. 000002, .x-"', =.000014. Hence, as Xp ^ a? ] -j- x' J -)- a; ^ , Xg ^^ a? 2 + -^2 + "^ s Xp=.0007U, a;s=— .000279, Xp +.a', r=. 000465. In Section I., § 23, it was shown that R, the drop per cent., can be expressed in the form R=lOo|.TSin0+/CosF is nearly the same for both [27r . 625 as against 27r . 600]. Regulation of equivalent Westinghouse Transformer. Power P\ac'tor Drop per cent. 1.0 ... 1.65 .9 ... 2.45 .8 ... 2.65 .6 ... 2.80 53. In order to obtain approximate values for the leakage coefficients of transformers of the core type it will be sufficiently accurate to consider the core as straight, and connecting two large masses of iron. Let the core be circular in section of radius r having coil 2 lying betwen the cylinders whose radii are r and r-\-l/.^ and coil 1, 58 Proceedings of the Royal Society of Victoria. lying between the cylinders whose radii are r+^-2 and r-^d.^-^-Zf^, and let r-\-i?.^ + ^i=rQ. The flux density due to Ci at all points on the cylinder of radius fg — z is iirnjCi z where A' is the length of the windings parallel to the core, and the flux in the Cj space between the cylinders of radii r^ — z and ^0 — (2 + <2^0) is = -^ fMr.-z)dz, and it is looped on that part that lies without it, hence it contributes energy ^E where Integrating between s^O and z^bi A' '(t-B ~' A' i 3 ■*"12)' as rQ^:.r-\-o-^-\-b.,. The current Cj also sends through the space occupied by coil 2 a uniform flux =^.|(.+AJ^-^} which is looped on all of Jifii and therefore contributes energy to the amount '\rb.^\, A' so that the energy of Cj due to those lines that it produces and that do not traverse the core is t it is also ^^L^Ci"^ nee Alternate Current Transformer. 59 In a similar manner we find that •VT AT LJ,' ^2, J^ I ■L*Al2— iU.21— y ( 2 6 j' but .•..'ri=^,-n-g + ^jr+ 1^ |-, and similarly From which we see that the leakage coefficient, x.,, of the coil next the core is negative. The equation giving .To + x-j can be written x, + x,=^^{b, + b.:) (Si + S,) (I.) where Sj and S.^ are the mean radii of the two coils ; and if ^^ and q.y be their space factors, their copper sections per unit length of winding are ^1^1 and ^2^2- These will be equal or very nearly so, and let l)^q.^=biq.,=s, then bi-\-b.i-='2sjQ, where Q==-^i^-, the harmonic mean of q^ and q.,. The total copper volume = 27rA'(Si^/i + S2^.A)=27rX'^(Si + S2)=7rQA'(^ + A,)(Si + S2), Volume of iron and o-=47r/u, t^ where A=:length of magnetic circuit, hence K^ volume of copper " - 3/xQX'- volume of iron ^ and if the transformer is being designed so that the copper losses are to be z times the iron losses at full load, then 60 Proceedings of the Royal Society of Victoria. Vol. copper I Vol. iron '^K where K and I are these losses per cm.'' •*• "^+"^=3;|r^ 4 ^"^-^ a form very suitable for the determination of x^-\-x.^ for core transformers. For example, if s = l, 1 = 10% K = 15.10^ <7i=.5, ^.3=. 7, Q = .58.3 as before, x■^-\-x^ for any core transformer designed on these data is given by X, + .T.,= - — ,,1.14. For a core transformer of the H type, simply wound, in which the rectangular opening in the stampings is 10.2x25.2 cm., and the width of the surrounding iron strip 8.9 cm. (See § 57) A' = 2x 2.5.2 = 50.4 cm. A=106.4cm. and if /x.= 2250, cKi+X2=.00075. 54. If a core transformer be wound with 2/ layers, i each of primary and secondary arranged alternately, and if D be the total depth of the windings, it can be shown that 4- ■^T) aji+a.2= -— ; — ^ X sum of the mean radii of all the layers, ^-— — ,^ X mean of the mean radii of all the layers, 3 A cri' which by exactly similar reasoning to that in § 5,3 can be put into either of the forms, .... 1 A'^ volume of copper P 3/aQA''^ volume of iron 1 X' si or x-.-\-x^= - '^ ' P S/xX" QK If there be / layers of one coil ajid / + 1 of the other, then we may take ,1 X" zl '^ ' /(/+1) 'iy.X" QK This result and those in § 53 will be sufficiently accurate for all practical purposes when the coils are rectangular in plan. Alternate Current Transformer. 61 The Transformer Numerics. 55. The numeric t { = Tj, = t., (q.p.) ) for transformers of any given type can be expressed in terms of the full-load output, periodicity, and tli.e magnetic and electric qualities of the iron and coppei-. Let us consider the case of transformers of the shell type similar to the one designed in Section II., with square windows (2/>,2i>), and iron tongue of square cross section (2^,2/3). From ;^ 33 r=A'^ ^^ (I.) where A is a constant depending on the iron and copper space factors. From the solution, as in § 32 of the equation which expresses the relation between the iron and copper losses at full load, we get /3-B^ (II.) in which B will be a constant, if, for all ti-ansformers of the series QK be constant. We may consider /> the iron space factor as fixed, and, provided the primary and secondary pressures remain the same, Q, the harmonic mean of the copper space factors, also as fixed ; and the above expression will be constant if z, the ratio of the copper to the iron losses at full load, be the same for all trans- formers of the series as well as the ratio K/I of copper to iron loss per cm^ at full load, both however, diminishing slightly in the same proportion as the capacity increases ; or s=Const. K = Ko(l -wP,) I=Io(l -;«P,). (HI.) where //i is a small fraction. Another way in which Is/K would be constant, and one more in accordance with the practice of some manufacturers, would be for K and Is each to be constant, I diminishing as the capacity increased, and z increasing in the same ratio ; or 62 Proceedings of the Royal Society of Victoria. K=Const., 1 = ^— ^, c = l+;;P.,. (IV.) where n is a small fraction. Again we have the full load output P,= i7fwAF (q.p.) for a non-inductive load on which the transformer would be rated, but and ^=W7=4//3'V|S^8- hence W -^' pSinS where D is a constant. Substituting in equation I. for b and ^, their values determined from II. and V., we get where M is a constant. Now I find for the same sample of iron that I is very nearly constant when w is constant over the range of flux densities, or of Is, commonly used in transformers, and that it increases slightly as iv diminishes. Taking it as constant, we get Hence if r for a transformer of a given type be known, the equation T" — ^= Const., will enable us to obtain fairly approximate values of t for other transformers of the same type that differ in capacity and periodicity. It is worth noting that equation V. above shows that, for equal heating or equal iron and copper losses per unit volume, the out- Alternate Current Transformer. 63 put of a transformer is proportional to 'V ^^—r- This is not proportional to the square root of w or of the frequency as, when w increases, fx for the same flux density will diminish and SinS will increase. Most Efficient Shapes of Transformers. 56. It has been shown (§ 21) that when consideration of leak- age is neglected, the measure of excellence of a transformer is T ShTS' hence the most efficient transformer of a given type and capacity and made of similar iron will be that one for which t is a maximum. If a, a be the total cross sections (insulation, etc., included) of the copper and iron circuits respectively, and /, A. their mean lengths, then -/X=^' al pi a/V "^ Qk = ^1 where g and z' are constants, Hence, as T for T to be a maximum, — and - a a must both be minima, and as the output where h is a constant, as the flux and current densities will be fixed, the problem resolves itself into finding values for the dimensions of the carcass that will make ^ , ^ , , • - — and - botli minima when aa is constant. a a Specifying the dimensions of a shell transformer in the usual way (window = 26, 26', tongue = 2y8, 2;S'), a = Ub', /=4(^ + ^' + 2^), a=4/5^', X=i{b+b'+(i). and proceeding by the method of indeterminate multipliers (A, B, C), 64 Proceedings of the Royal Society of Victoria. in which the coefficients of db, db\ d^, and dji' being equated to zero give us, Eliminating A, B, and C from any two sets of three of these equations we get the two relations which show that b'>'2b and 2;8and<3/3. Letb' = $h, f3' = i](S, p=ub, and equations I. can be put in the forms U=2};-^^-^ j (II. ?/ + 2' '^~ ti + '. Su + i 2// +31" by means of which the equation of the losses a/ fil becomes Cin + 4)(3//^ + &u + 2) ^^ /I jjj ?/-^(2?^ + 3)(/r + 6?^ + 6) ""QK' ^ '' from which « (the one positive root) can be determined by trial when s/I/QK is known. ^ and r] are found from ti by equations IL, and so the shapes of window and tongue and their relative sizes are determined. The i-elation | wn.,C^=^^., can now be reduced to Altei'iiate Current Trans former. 65 from which /», and hence the transformer, is determined. The equation for t can be put in the form _ TT/i-Pa 1 orr=''^ 1^ (V.) by means of which it can be quickly calculated, and it will be found that the result is a true maximum. For example, assuming the same data for design as are adopted in §§41 and 52, .|-^= 1.029, and equation III. gives u=l.l, hence by means of II. we find that ^' = 2.35^, (3' = 2A8/3, which with /3=l.lb, give the most etficient shape for a shell transformer in which z/I/QK= 1.029. If P.2 = 12.5 K. W., the same capacity as that of the transformers in § 41, equation IV gives 6=4.55, and equation V., T=7300. The losses being a/ QK and aA/I, we find that each is equal to 181 watts, so that the efficiency at full load is 97.2 per cent. This maximum efficiency transformer will not have such good regulation on inductive loads as others less efficient, but with relatively wider windows. A compromise between efficiency and regulation can always be made suitable to the nature of the work the transformer is intended for. For the above transformer, if wound in five sections, Xi + X2:= .00075; and the regulation would be, for a non-inductive load, 1.55 per cent., and for an inductive load of .8 power factor, 3.7 per cent. These figures can be compared with those in § 52. 57. A core transformer of the H type, in which the magnetic circuit is rectangular (2(3, 2/?') in section and the coils rect- angular in plan, is exactly the same in geometrical shape as a shell transformer, but the copper and iron circuits of the former occupy the places of the iron and copper circuits of the latter. 66 Proceedings of the Royal Society of Victoria. Let 26, 26' be the dimensions of the rectangular windows, or winding apertures in the laminae, the coils being wound round the lb' dimension, 2/3 the width of the iron strip, and 2/3' the dimension of the core measured perpendicular to the laminae, then a = ibb', /=4:{P + (3' + b), a = 4/3^', \ = 4:{d + b' + 2(3). and we find as in § 56, or by simply interchanging /3 and b, /3' and b' in I., § 56, that for maximum r, that is maximum efficiency 6(3^-yS') = 2/3(/3'-2/3), b{b'-2b)=p{3b-b'). If b' = ib, fi' = rj^, and /S^z/i as before, • 3u + 2 iu + 3 ^=17TT' ^ = 2^7:^' ^"-^ and the equation of the losses is (3?/ + 2)(6?/''' + 6?/ + l) /I provided the coils are wound in a number of alternate layers so that the mean lengths of the primary and secondary turns are equal. From this equation u can be found, and thence by II., $ and rj. The equation of the output (see § 56, IV.) gives h, which with ?/, ^ and rj, determine the transformer. In this case ^ tt/aP^ 1 2pC^yV' ( 1+ ^ + 2?^) ( 1 + ?/ + llrj) For example, if P,= 12.5 K.W. as before, then ?^ = . 876, ^=2.47, -7=2.36, 6 = 5.1, and T=7320, just the least thing better than the maximum efficiency transformer of the shell type. Alternate Current Transformer. 67 If s/I/QK=:l, max. T would be the same for both types, and if 2^I/QK<1, the shell type would be the better. Magnetic leakage is in general less, and good regulation more easy to attain in core transformers than in shell transformers. To enable a comparison to be made with the shell transformer in the last paragraph, we will determine the sum of the leakage coefficients and the regulation for different kinds of load of the core transformer considered above, supposing it to be wound (rt) in thi'ee layers, one primary and two secondary or vice versa ; (h) in five layers, two primary and three secondary or vice versa. From § 54, 1 2^ ^ ij 3/.X'2 QK' and A=4(^ + // + 2^). A' = 45', so that, using the same values for the constants as before, we find, for (a) A-i + .T2 = . 000381, (h) a"i+X2 = .000127, from which, proceeding as in § 52, we find for the regulation A'l + X^ Drop per cent. Power Factor. («) Three layers. Five la.ver.s. 1.0 .8 .6 1.49 2.43 2.55 1.47 1.58 1.43 58. It is obvious that in core transformers of the ring type in which the winding is continuous all round, the maximum efficiency shape will, other things being equal, be that in which the magnetic circuit is shortest, that is when the opening in the laminae is filled with the copper circuits. The ring type is not suitable for practical construction, but a near approach to it is the Burnand transformer,* in which the magnetic circuit is formed of square laminae from which a symmetrically placed inner square has been removed to give the winding space. Each side of the square is built and wound separately with triangular 'See "Electrician," Sept. 19, 1902. 68 Proceedings of the Royal Society of Victoria. shaped windings, and the four sides jointed together to form the completed transformer. Let us determine the proportions of such a transformer so that T, and hence the efficiency, shall be a maximum. Let 26, 26, be the square opening in the laminae, 2yS, 2^' the cross section of the magnetic circuit, 2/3 being measured in the planes of the laminae, then, « = 46^ /=4(/3+/3' + i6) a = 4:^/3', \=8{b+/3) Proceeding as in § 56 we find, in order that A / , — and — shall be minima a a when aa is constant, that 2/3(/3'-2/3) = 6(3/3-;8') and if p' = -qjS, p = ub as before, j^3— 7/ _4/^4-3 and the equation of the losses ai _ pi becomes 18?/ +16// + 2 _^pl 6?/(4«^ + 7?/+~3)~^QK' from which, for any given values of s, />, Q, I and K, tc can be found and hence rj. The equation of the output, 6V^=. ^-^ , gives b, which with ii and 77, determine the transformer. For example, if we take as before s^ = 1.029, P,= 12.5 K.W. we find b=7.S, /3 = 4.5, /3'=11.09, and the value of r is 7680, which is considerably larger and hence better than for either of the two preceding types. Iron loss:=:copper loss = 176.7 watts. Efficiency = 97.26 per cent. Alternate Current Transformer. 69 These transformers are wound in five or seven layers and their regulation is of a very high order. The formula in § 54 would only enable us to obtain a very rough approximation to x-^ + x^ for this type. General Solution of the Transformer Problem by a Vector Method. Explanatory, 59. {a) If a be any vector representing e.m.f., current, or flux, on the plane alternate current diagram (Fig. 2) and if we under- stand by (.a the vector got by rotating a through a right angle in the positive direction, and hence if we understand by (Oos^ + tSin^)aor e'-^o. the vector got by rotating a through the angle Q in the positive direction, then it is well-known that operators such as e^^ can be manipulated as ordinary algebraic symbols, and that t can be treated as if it were the algebraic imaginary \/ _ \* (b) If rtj, a.2, «3 etc., be numerical multipliers, then the vector or the resultant or sum of the vectors a^e'-^ia, a,/-^-^a, a^e'-^^a, etc. is ={2aCos(9 + t2aSin6'}a = A(Cosi/^-|-tSini/^)a=Ae'''/'a where A2 = (2rtCos^f + (2aSin^f = ^a^+'2%a^a.,Go^{d^-e.) and hence the operator fli^'^i + a./'^2 -1- rtge'^s + etc. = A^"A where A and i/^ are given by the above equations. *Lyle. Alternate Current Prohleim. "Electrician," 41, pp. 816-818; 42, pp. 72-74 and 148-151, 1898. 70 Proceedings of the Royal Society of Victoria. (c) If a represent the harmonically varying quantity nCos7vf, then since ■T, (nCos7Ci/)=.zvnCoslzvf + :^ j TT we "^ a or wia will represent — (;/Cosz£'/), and we may write dt IT d '2 —az=we a = 2VLa. at 60. If o-/47r be the permeance of the magnetic circuit, closed or open, and limited in section by the iron core where the latter exists ; and if 8 be the angle of magnetic lag of the iron, then as the flux density remains very nearly constant throughout the range of operation of a transformer, we may without much error consider o- and 8 as constants. The total number Nj of magnetic lines looped on the Wj turns of the primary coil is the sum of three sets, namely, 1. Those traversing the iron core, produced by the magnetising ampere turns Jl^G■^-\- n.^Cj, and behind them in phase by the angle 8. Hence these = (re («iCi + n.J^.^ 2. Those produced by C^ and in phase with it that miss the iron core. Let these 3. Those produced by C.2 and in phase with it that miss the iron core. Let these Hence similarly where x^.^ and x-^.^ have similar significations with regard to the secondary coil that x^^ and x^i have with regard to the primary. Alternate Current Transformer. 7t We thus have four leakage coefficients and it will be noticed that they are connected with the two coefficients x^ and x.^ hitherto used by the equations X., = x.2.2 - .r,2 (see ^ 43). 61. The equations of motion are El = rfi, + nij^^i = nCi + wn/'^^ Nj (I- ) E.,= - r.Co-fio :i,No = - r..C.,-wn./^ K, (H-) -- ■' at " "^ where Ej, E.^, are the terminal e.m.f's, and r^, r^, the internal resistances of the coils. If R be the external resistance or its equivalent in the secondary circuit, and Cos^ the power-factor of the load, E.,Cos<^ = r/'^C, (III.) Eliminating E.^ between equations (II.) and (III.) and putting = Ti -— = T.i, R -Coscfi=e [Note that the 6 here is the same as the ^Cos^ in the early part of this papei\] we get {l+x,jS;fi, = -(l + xjKle- + 9^\ + ] , X,^'= 1 + 2.r,,Cos8 + x,i ; and that 72 Proceedings of the Royal Society of Victoria. Cos(S + <^) + a-i2Cos,)Sin8 + ^ + ^^ j tan/3= ^ " " ^fe^^ Sin(S + c^) + .^128111^ +^|l +(^i2 + X22)Cos8+ -2^-{-Xi^xA where it — /? is the angle that Cj is behind C^ in phase. 62. Eliminating Cg from equations (I.) and (IV.) and putting X-^i X^i ~r •^22 "^12 ^^ -^) 1 _ ■^11^22 ^12"^21 ^'j T1T2 — + — ^ = «, To T, we get «lJ^l_ Tl g ^2 +6'-^ +ir22 + — ^ '^ T2 + _L_ ^ ^ „^Ci e ^2 ^'+d)e +x.„ + — e 9 f from which by § 59, b, we find that A ^iTi ' D ^ "^ where D^=l + 2.riiCosS + 2 — +V+ -, + 2^(xSin(^ + TCos<^ Tl Ti I + (x-iiX + -)Sin(8 + <^) + (a;uT--)Cos(S + <^) + «Cos(S-<^) — wSin(8 - ^) + r«ii;« + — jSint^-hTiCi] « — — ^Cos^ I also, if a be the angle that C, is behind E^ in phase, so that Cosa is the power factor of the transformer, Alternate Current T^^ans former. 73 D A Cosa= SinS + - + e-{cos

rX.JCos{h + +-^ )Sin8 + 2— Sin0 + (jCi.,x-,,i + ~ )Cos0| \ T2 '''1 ^ ''"l''"2 / r^i^22 ■ ■^21 ~l" ■^12~| p Sil'E^'^ t ^ ■'^12^21^ L Tj r2 J Tl TiTa^ T.^ 3 Q=^„ + ^,^ + ^#(say). (VII.) The power Pj taken in by tlie transformer on the primary side being = lEiCiCosa we find P=l?J % (VIII.) 63. From equations (V.) and (VI.) we get ex,., r,r, D R ;72C,_«iEi 1 ^jj^> and as E2Cos<^=RC2 and 0= j^ Cos^, we find that E2='1^^^E,. (X.) As the output P2 = ^E2C„Cos<^ we find that, substituting for E2 and C2, that P,=^3' ^'^Oos.^. (XI.) 64. Equation (IV.) of i^ 61 can be written in the form, (1 +*•!./) {n,C, + n,C,) = - {(x^-x,^e'^+^e " '(l~V + ^ T2 but niC^ + n.jG^i—F/a; and, by § 59, fi, we find that F/o- «A (XII.) 74 Proceedings of the Royal Society of Victoria. where ]\r =1+2^ |(,x%,-.>v.)Sin and its maximum value is Xig'^Cosc^ ?i + 2v/^of2 67. Thus, without making any assumptions as i-egards leakage, all the important variables in the general theory of the trans- Proc. R.S. Victoria, 1904. Plate 1. 15000 5^ . Proc. R.8. Victuria, 1904. Plate II. Fig. 2. 0 A" Proc. R.S. Vidtiria, 1004. Plate III. Fis.^. Proc. R.S. Vidoria, 1904. Plate IV. Fi£t. 5. s I'lii'. "11 f D ~- - ^ / i A ■:■:-:& y-.-:. '•I'ii - z ^^ .::.p.\ .. } 3' p ^SOpoAaq ■e iA/nu> dtie-yto C, | Fig. 6. 9l ;a t:::^, s Proc. R.S. Victoria, 1904. Plate V. Fig. 7. /y 5' 1 r 0 /I 2 5 4 1 M.M.F. diaauim 0. / / / p 1 .1 1 ' / Alternate Current Transformer. 75 former liave been expressed in terms of 0 l)y equations identical in form to those obtained in Section I., and which can be reduced to the latter by making Xi.i^x.2i=^0 and dropping insignificant terms. From equation (XI.), § 63, we can, as in § 18, express $ in a series of ascending powers of Pa, and thence transform the pre ceding equations, in which the independent variable is 6, to others in which the independent varialile will be P.^ or the output. This transformation, and any further discussion of the general equations we have obtained is unnecessary, as it would follow on exactly similar lines to what has been already given in Section I. Art. II. — Contributions to our Knowledge of the Anatomy of Notorydes typhlojps, Stirling. Parts I. and II. By GEORGINA SWEET, D.Sc, Melbourne University. (Communicated by Professor W. Baldwin Spencer, F.R.S., &c.). (With Plates VI-IX.). [Eead 12th May, 1904.] Introduction. The investigation, of which the following forms a record, has been carried on in the Biological Laboratory of the Melbourne University, for the use of which I have to thank Professor Spencer, who has also very generously placed his splendid stock of animals at my disposal, and has given me facilities in obtain- ing literature, some of which I might otherwise not have seen. The subject matter falls naturally into three parts, each of which is complete in itself, though they are to a certain extent interrelated. Part III., on the Eye, is now ready for the press, an abstract of it having been read at the Dunedin Meeting of the Australasian Association for the Advancement of Science, in January, 1904. Part I. — Nose, with Organ of Jacobson and Associated Parts. Of the various structures to which of more recent years con- siderable attention has been directed, not the least interesting is the Organ of Jacobson, and with it the relations of the cartilages and bones of this region. Especially is this so in view of the valuable papers by Dr. Broom, on its comparative anatomy in the various groups of the Metatheria and Eutheria, in which he Anatomy of Notorydes typJdops. 77 claims that, on account of the very slight tendency of these parts to vary with external variations, " we have a factor of considerable value in the classification of the Eutheria, probably of more value than eitlier dentition or placeutution." If this be so, and there seems strong evidence in its favour, we ought to tind in this organ data on which to base a true conception of the relationships of such an aberrant form as Notoryctes, especially valuable since its embryology remains at present unknown. In itself a desirable result, this should also assist in defining the affinities of associated groups. Heretofore, apparently, nothing has been known of its structure in Notoryctes, nor even of its presence. This being so, it was suggested when working out the relations of the naso-lachryraal duct in connection with the eye, that I should include Jacobson's Organ in this research. Further- more, we find that the structure and relations of the cartilages and bones associated with the nose are well worthy of record, as well as those of the organ itself. In Broom's valuable thesis on Jacobson's Organ, ^ he has dis- tinguished four types of this structure in mammals corresponding in part to the main groups : Monotreme, the most highly developed, Marsupial, Rodent, and general Eutherian ; the main features of distinction being the character of the connection of the lumen of Jacobson's Organ with the naso-palatine canal, or with the nasal cavity, and tlie arrangement and degree of complexity of the cartilages. In view of the apparent value of this organ in classification, and the much modified character of Notoryctes, I have thought it desirable to make a more complete comparison of the various details of structure, with similar parts in other forms, than might have been necessary in some other animals. Posit loti of the Organ of Jacobson. The organs of Jacobson are, as stated above, well developed in Notoryctes, being approximately equal in size to those of the Rabbit. They are situated near the floor of the nasal cavity, one on either side of the median line, just in front of the vertical plane of the osseous nasal septum, i.e., 3.6 to 3.8 mm. from the anterior edge of the snout. They are separated from each other 1 Trans. Roy. Soo. Edin., vol. xxxi.x., 1898-1900, p. 234. 78 Proceedimjs of the Royal Society of Victoria. and partly enclosed by the bony palatine processes of the pre- maxillary bones (the prevoniers of Broom) and by the cartilages of Jacobson. The Organ, with a small ledge of cartilage lying externally to it, causes an elongated triangular projection (Figs. 3, 4, 5, i.s.r) on the mesial wall of the nasal furrow, the base of the triangles being formed by the lateral wall of Jacobson's Organ. This ridge in the lining mucous membrane of the nasal furrow, which is always indicative in mammals of the position of this Organ, has been called by Broom "the inferior septal ridge." It is continued in a less degree anteriorly and posteriorly ; anteriorly because of the presence of the cartilaginous shelf supporting the Organ, in front of the Organ itself, the trough so caused being here occupied by glands ; while posteriorly the lower part of the ridge is still present, because of the bony shelf from the palatine processes of the premaxillary bones. Even where the Organ of Jacobson is itself present, the size of its consequent ridge is increased by a considerable development of glandular alveoli, outside Jacobson's cartilage and continuous with the gland masses in front of and behind Jacobson's Organ — compare Phascogale^ Didelphys,'^ Perameles,'^ Pseudochirus.^ Cartilages and Bones in connection with the Nasal Organ. The cartilaginous nasal septum (Figs. 1 and 2, n.s.c.) is present, dividing the nostrils right up to the anterior end of the snout. Its cartilage is hyaline, and the cells numerous, deeply staining and showing evidence of rapid growth. As have others, we find that transverse vertical sections offer the best means of studying this part, aided also by longitudinal vertical sections. Beginning anteriorly, we find that the alinasal cartilage supporting each nostril is well developed, and is free anteriorly on its lower border (Fig. 1, a.c), not being here united to the ventral processes of the septum, but swelling out instead into an edge which is club-shaped in transverse section, and supports a well marked ridge [p.l.r.), the cartilage being covered with a considerable thickness of gland material (m.g.), the whole rendering the cavity of the nostril crescentic in outline. 1 Broom : Proc. Linn. Soc, N.S.W., vol. xxi., 1W96, p. 593. 2 Loc. cit., p. 597. 3 Loc. cit., p. 599-000. 4 Loc. cit., p. 603. Anatomy of Notoryctcs typhlops. 79 In outline the cartilages present in transverse section, that of an ornamental T. This ridge is referred to by Dr. Stirling^, and is well shown in his accompanying figure of the animal. Dorsal to these alinasals lie the forward processes of the nasal bones (n.d.). Further back, the ridge containing the swollen edge of the alinasals, comes to lie more ventralwards, the superior position being taken by another ridge (Fig. 5, s./.r.) with glandular interior, and containing a large duct from the mucous glands posterior to this level. A short distance behind the beginning of the second ridge there arises a process from the arch of the ali- nasal cartilage on each side, connecting each with one of the ventral processes of the nasal septum (n./.r.), so that at this point, and not anterior to it, the nostril is completely enclosed on each side with cartilage. This condition is closely comparable with that in Macro.scelides, as shown in the figures given by Broom'^. Soon there pierces the alinasal cartilages a canal on each side, through which passes one of the naso-lachrymal ducts, which open in front of this, on the ventral surface of the primary lateral ridge into the ventral nasal furrow [v.n./.) on each side. Between the plane of the opening of the naso-lachrymal duct, and that of its passage through the encircling cartilage, I have heen able to trace a splitting off from the ventral surface of the carti- lage of the nasal floor (formed by ventral processes from the carti- laginous septum), of what is at first a thin lamella of cartilage, in three parts. Those on either side lose their connections with the nasal floor except for a while at the extreme outer edge of each, and finally become continuous with the anterior edge of each premaxillary bone. The central portion remains longer in con- nection with the nasal cartilage, so that, in a transverse section taken just at the level of the passage of the naso-lachrymal ducts through the alinasal cartilages, the following relations exist. The nasal septum (Fig. 1, n.s.c.) is very thin and deep, giving off above the two alinasal cartilages {a.c), and ventrally two pro- cesses [n.f.c.) forming the floor of the nasal cavity. Compare in this respect Ornithorhynchus, in which the nasal septum becomes united with these nasal floor cartilajres." This is to be 1 Stirling: Trans. Roy. Soc. S. Aus., 1891, p. 159, pi. iii. 2 Proc. Zool. Soc. Lond, 1902, vol. i., pi. x.vi., fig. 1. 3 Broom: Trans. Roy. Soc. Edin., vol. xxxix., p. 235. 80 Proceedings of the Royal Society of Victoria. contrasted with the condition found in most Marsupials, and also in Rodents, in which they are at most in contact with the nasal septum. In the Macropodidae, Symington^ has noted the con- nection of the nasal floor cartilages to the ventral edge of the nasal septum, by perichondrium. Beneath this, and separated from the cartilage by connective tissue and blood-vessels, are the extreme anterior ends of the premaxillary bones {p.b ), just losing their fibrous cartilaginous connections with the lateral edges of the nasal floor, and separated from each other in the middle line by a large vein (v.). Above this vein is a somewhat wedge-shaped nodule of hyaline cartilage {p-c), which in the next section posteriorly sends down a fibrous process to occupy the space between the two premaxil- laries, pushing the vein ventralwards. In this section, also, the nasal bones (n.l>.) have grown down, enclosing the alinasal cartilages nearly to the level of the primary ridge, and three sections further back the nasal and upward processes of the premaxillary bones meet, completing the bony as well as the cartilaginous capsules round the nose. Still proceeding back- wards, we find that the wedge-shaped cartilage has now com- pletely descended between the premaxillaries to form the connection between their mesial edges. The above description can be readily corroborated on reference to longitudinal sections. Splitting off anteriorly from the ventral edge of the septum is the narrow sheet of fibrous cartilage passing obliquely downwards and backwards to lie between the palatal processes of the pre- maxillary bones in their anterior part. On the hinder face of this sheet of cartilage is the hyaline cartilaginous swelling, which in transverse section appears wedge-shaped. In front of the sheet, the bones are separated by a well-defined vein, connected with a large blood sinus, which curves round vertically in front of the cartilaginous septum. Posteriorly in these longitudinal sections we can see that the central cartilaginous bar or narrow sheet becomes lost as the two palatal processes of the premaxillae become more intimately united. There can be, I think, no doubt but that this central cartilage repre- sents here the prenasal cartilage of other animals. Its general 1 Jour. Anat. and Phys., vol. 26, p. 372, and pi. x., fig. 1. Anatomy of Notoryctes typhlops. 81 relations greatly resemble those shown by Broom to exist in the foetal calf.^ With reference to the transverse plates of cartilage described by Broom as existing on each of the central rod, and supporting the papilla between the naso-palatine canals, which is so marked in Marsupials,^ such for example as in Didelphys murina,'' in Perameles nasuta/ in Petaurus,'^ and Trichosurus,*^ Phascolomys'' and Macropus.** I can find no trace of hyaline cartilage in such a position, but the fibrous sheet of cartilage which connects the main part of this prenasal between the premaxillary processes with the nasal septum, sends out laterally a thin ill-defined fibrous layer (Figs. 2 and 3, f.p.c), which extends backwards beneath the palatal processes into the papilla, behind which it does not exist. Apparently this represents the papillary cartilage of other Marsupials, and that of Miniopterus" and Macroscelides.^" Returning to the vertical transverse sections, we find that not only the primary and secondary lateral ridges, but also the septal cartilage are covered by a great thickness of glandular alveoli, forming on the septum the superior septal ridge {s.s.r.). These glands have well defined ducts, often. 06 mm. in diameter, running longitudinally, to open far forwards into the vestibule. The thickness of the glandular layer varies on the superior septal ridge .24: to .52 mm., and on the superior lateral ridge .24 to .6 mm. The lining membrane of the nasal cavity over these i-idges is smooth, like that of the Guinea-pig, and so unlike that of the Rabbit, which is much plicated. About this vertical plane, the cartilaginous projection, supporting the primary ridge from the lateral wall, diminishes greatly in size and finally disappears, so that on each side the cartilages of the nasal floor now form a very shallow double U-shaped curve, each of the nasal furrows of each side occupying the loop of one U, the mesial edge of the 1 Proc. Linn. Soc. N.S.W., vol. x., n.s., pi. xliv., fig. 7, and p. 561. 2 Loc. cit., fitf. 6, and p. 560. 3 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1896, p. 597. 4 Loc. cit., p. 599. 5 Loe. cit., p. 604. 6 Loc. cit., p. 607. 7 Loc. cit., p. 613. 8 Loc. cit., p. 610. 9 Loc. cit., vol. X., n.s., 1895, pi. xliv., figs. 4, 5, p. 560. 10 Proc. Zool. Soc. Lond., 1902, vol. i., pi. xxi., figs. 8, 10, p. 226. 82 Proceedings of the Royal Society of Victoria. inner U being continuous with the ventral edge of the nasal septum, while the naso-lachrymal duct lies underneath in the angle formed between the two loops. The arrangement of this double U-shaped cartilage, and the subsequent reduction of the cartilages (to be immediately described in Notoryctes), may be compared with that shown in Klein's figures of the Guinea-pig,^ though here the central cartilage is not connected with the septum as it is in Notoryctes. Gradually here the outer U becomes lost on each side at about the level of Stenson's duct, or a little posterior to that duct. Compare this with Didelphys murina,^ Perameles,^ Aepyprymnus,^ and contrast with Tricho- surus.® In Notoryctes, however, this outer cartilage is present behind the naso-palatine canal as a rudiment. At the same time, the alinasal cartilages, having receded dorsally, only extend down in the upper third of the nasal wall. At this point, each nasal cavity (Fig. 2, n.c.) in transverse section resembles a two- pronged fork, the two prongs being represented by the two nasal furrows (v.n./.). A change is also noticeable in the outlines of the palatal processes of the premaxillae, which are still united only by the median cartilage above described. The adjacent edges of the premaxillaries, which are thin anteriorly, become much thickened posteriorly (Fig. 2, p.p-p.\ wedge-shaped in cross section, their mesial faces being convex to each other. The upper edge of this wedge now becomes more marked, rising up in a crescentic fashion (compare especially Phascolgale^, in which, however, they are much smaller than in Notoryctes and Perameles^), till it touches the ventral cartilage of the nasal floor, the lower edge of the wedge disappearing. In this plane (Fig. 2, J.c.) there appears a swelling in the hyaline cartilage of the nasal floor, from which passes back a bar, also of hyaline cartilage, part of Jacobson's cartilage. The swelling in the mesial wall of the ventral nasal furrow, i.e., the inferior septal ridge (i.s.r.), caused by the cartilage, 1 Quart. Jour. Micro. Science, vol. xxi., pi. xvi., fig. 1, 2, 3. 2 Proc. Linn. Soc. N.S.W., vol. si., n.a., 1896, p. 597. 3 Loc. cit., p. 601. 4 Loc. cit., p. 610. 5 Loc. cit., p.^607. 6 Loc. cit., p. 593. 7 Loc. cit., p. 599. Anatomy of Notary ctes typhlops. 83 and which increases in size greatly and almost immediately, occasions a pushing in of the ventral nasal furrow, the cavity of which now becomes in cross-section foot-shaped, the inferior septal ridge filling up the instep. Suddenly, just posterior to this, there appears the swollen anterior end of Jacobson's Organ (Fig. 3, J.O.^. The cartilage of the nasal floor may be now called in part Jacobson's cartilage, since it has here lost its connection with the nasal septum (Fig. 3, /.c^j. In this respect Notoryctes resembles Ornithorhynchus^ and Echidna,* in which Jacobson's cartilage "is continuous in front of the naso-palatine foramen with the cartilage in the floor of the nose," as also with the septum, "while behind it is separate." It resembles also the Rabbit,^ and also the Guinea-pig,'' in that the cartilage is con- tinuous with the cartilage of the nasal floor, though in each of the latter the cartilage of Jacobson is altogether independent of the cartilaginous nasal septum. The cartilage of Jacobson now consists, on each side, of a crescentic shelf, from the middle of the concavity of which rises, at right angles, a band of cartilage (o.J.c), under which runs, near its anterior end, Jacobson's duct (Fig. 3, J-d.) into the "toe" of the nasal furrow, while in the groove formed between the band and the upper horn of the cres- cent lies the Organ of Jacobson. In Notoryctes, the crescentic cartilage of Jacobson is oblique, similar to that of Petaurus*, and unlike that of Pseudochirus and Petauroides", which are more vertical. The band or shelf of cartilage supporting the lateral wall of Jacobson's Organ, is comparable in part to what is called the septal turbinal in Macroscelides,' though arising from the main cartilage at a diflTerent angle. It is further comparable to Macroscelides in that this shelf is only connected with the ventral cartilage behind the exit of Jacobson's duct from the Organ, near its anterior end. This outer bar is similarly found in most Marsupials, but that in Notoryctes differs from them in 1 Proc. Zool. Soc, 1891, p. 578. 2 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1896, p. 592. 3 Q.J.M.S., vol. xxi., p. 550. 4 Loc. cit., p. 220. 5 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1896, p. 004. 6 Loc. cit., p. 604. 7 Proc. Zool. Soc, vol. i., pi. xxi., figs. 3 and 4, p. 226. 84 Proceedings of the Royal Society of Victoria. one particular, viz., that in them this bar is connected above and in front with the upper end of Jacobson's cartilage, and below and behind with its lower outer edge. In Petaurus,^ however, and Phalangers,^ and to a less extent in Trichosurus" and Macropods,^ there is a ridge process exactly similar to that of Notoryctes in its origin from the inner upper side of Jacobson's cartilage, becoming detached from it, and then more posteriorly becoming attached to the lower ridge of the cartilage. The " bar " in Notoryctes apparently truly corresponds to that of the other marsupials in that it comes off anteriorly to Jacobson's duct from the ridge process, curls round the Organ and over the duct, and becomes attached posteriorly to tlie duct, to the ventral edge of Jacobson's cartilage, being therefore merely a further exaggeration of what is present in Petaurus, and the Phalangers generally. Meanwhile in Notoryctes, the cartilaginous connection between the palatine processes of the premaxillaries has almost disappeared, the two bones by this time practically fusing. The crescentic character of the bones now harmonises closely with that of the cartilages (Figs. 3 and 4, p. p.p. andyit'.). From the ventral convex surface of the rapidly dwindling cartilage of the outer nasal tloor, is given out just here a small process of hyaline cartilage (Fig. 3, s.c), which is found strengthening the upper and anterior wall of Stenson's duct which lies just posterior to this. Here we have another point of difference from other Marsupials, in which there is no cartilaginous support to the naso-palatine canal, though in Petaurus^ and others we find a process supporting the inner wall. This may also be compared with the Rabbit," in which Stenson's cartilage is a continuation from the cartilage of the nasal floor, and contrasted with the Guinea-pig,^ in which the cartilage forms a closed capsule around the two ducts, and is quite separated from all other cartilages. It is to be noticed here, that the upper horn of the crescentic Jacobson's cartilage is 1 Proc. Limi. Soc. N.S.W., vol. xi. ii.s., 1896, p. 601, pi. xliv., fijfs. 10, 11. 2 hoc. cit., p. 616. 3 Loc. cit, p. 607. 4 Loc. cit., p. 618. 5 Broom : Trans. Roy. Soc. Edin., xxxix., p. 240. 6 Klein : Q.J. M.S., vol. xxi., p. 555. 7 Loc. cit., p. 228. Anatomy of Notovyctes typhlops. 85 thinning out greatly, as also the cartilage underlying tlie ventral nasal furrow, so that, about the level of the exit of Stensoii's duct from the nasal furrow, there is no cartilage left in this region, except for remnants of the outer nasal floor cartilage (Fig. 4, n.f.c), and the outer bar of Jacobson's cartilage {o./.b.). Thus, hei-e the median and lower lateral parts of the cartilaginous crescent disappear first as compared with the Rabbit,^ and, contrasted with the Guinea-pig,-^ the upper lateral or lower lateral parts of which go tirst. At first this remnant of cartilage appears to become directly connected by its perichondrium with the lower edge of the crescentic bone (Fig. 5) as found by Klein in the Guinea-pig ; soon the cartilage disappears altogether, leaving a very thin bony shelf (Fig. b, p.p.s.) in its place. Compare this with Perameles,^ and also with the Macropodidae* in so far that the cartilages of Jacobson form an incomplete tube, becoming reduced posteriorly. At this level, nerve fibres occupy almost the whole space between the bone and the mesial wall of the Organ. Posterior to the Organ of Jacobson the inferior septal ridge still remains because of the persistence of the bony shelf, which anteriorly helped to support Jacobson's Organ ; while, as far for- ward as the anterior end of the Organ, the primary lateral ridge {p.l.r.), which has been for a short distance devoid of .special support, is invaded by a thin lamina of bone from the maxillary bone, becoming the maxillo-turbinal {m.t.). In the hinder part of this region the palatal processes are overlain in the middle line by the anterior portion of the vomer, so that there is now a complete bony partition between the right and left nasal cavities, from dorsal to ventral or palatal surfaces. Ducts of Jacobson and of Stenson. The duct connecting the lumen of Jacobson's Organ with the nasal cavity (Fig. ?>,J.d.) is very short, .06 mm.., since the wall enclosing the ventral sulcus of the extreme anterior end of the Organ lies almost immediately in contact with the mesial edge, 1 Klein: Q.J.M.S., vol. xxi., p. 554. 2 Loc. cit., pi. vii., flg. 2. 3 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1896, p. 600, fig. 8. 4 Jour. Anat. and Phjs., vol. 26, p. 372. 86 Proceedings of the Royal Society of Victoria. which is also the most ventral part of the nasal furrow (v.n.f.). The duct then passes outwards almost horizontally to open into the nasal furrow. At this plane, in transverse sections, is also seen the external aperture of the naso-palatine, or Stenson's duct (Fig. 3, n./>.if.), into the mouth. This duct, which is .40 mm. long, runs inwards, upwards, and backwards, piercing between the premaxillae and palatine processes to its origin from the ventral edge of the nasal furrow, some distance behind the open- ing of Jacobson's duct into it (Fig. 4, n.p.d.). There is, therefore, no direct communication between the cavity of the Organ and Stenson's duct, except through the cavity of the nasal furrow ; this is confirmed by the difference in structure between the wall of Jacobson's duct and of Stenson's duct, and the intervening nasal furrow. This condition may be compared with that described by Broom as an exception among Marsupials in Aepyprymnus^ by Klein in the Guinea-pig^ and Rabbit,* by Harvey in the Rat and Hedgehog,* and by Broom in Dasypus.* It may also be contrasted with that in Ornithorhynchus" and Dog,'' and the usual Marsupial and higher Mammalian types, as described by Jacobson, Gratiolet, Balogh, Fleischer, and Broom, in which Jacobson's Organ opens into Stenson's duct, otherwise remaining closed, e.g.^ in Macro- pus*, Phascologale', Dasyurus^", Didelphys", Perameles^'*, and Phascolomys." The openings of Stenson's ducts into the mouth cavity are separated by a well-marked papilla, the centre of which becomes somewhat hollowed out (Fig. 3). This, as stated above, is supported, anteriorly at least, by a fibrous continuation from the prenasal cartilage {f.p.c). The similarity of the general relations of the parts seen in such a section of Aepyprymnus as 1 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1896, p. 610. •2 Q.J.M.S., vol. xxi., p. 219. 3 Loc. cit., p. 555-6. 4 Q.J.M.S., vol. xxii., p. 50. 5 Trana. Roy. Soc. Edin., vol. xxxix., p. 242. 6 Proc. Zool. Soc. Lond., 1891, p. 578. 7 Q.J.M.S., vol. xxii., p. 301-2. 8 Jour. Anat. and Phys., vol. xxvi., p. 372. 9 Proc. Linn. Soc. N.S.VV., vol. xi., n.s., 1896, p. 593. 10 Loe. cit., p. 594. 11 Loc. cit., p. 597. 12 Loc. eit., p. 600. 13 Loc. cit., p. 613. Anatomy of Notoryctes typhloj^s. 87 that shown by Broom^ to those seen in a similar section of Notoryctes is considerable, especially in reference to Jacobson's cartilage, the opening of the duct into the nasal cavity, and its relation in vertical plane to the dorsal opening of the naso- palatine duct into the nasal furrow, and to its ventral opening into the mouth. General Structure of Jacobson's Organ. As in the Organ of Jacobson previously described in other animals, the lumen of the tube (Figs. 3, 4, 5, y. (9.) in Notoryctes is more or less laterally compressed in its main portion, so that we distinguish the lateral {l.w.) and median walls (Fig. 6, m.w.), which meet at the upper and lower sulci. In the examples of which I have sections, the left tube is greater in vertical diameter than is the right, the latter, moreover, in great part of its length being almost circular, while, right to the hinder end, the left organ retains, in an increasingly marked manner, its compressed character, its cavity being posteriorly a mere slit. As usual, the sensory epithelium is confined more or less strictly to the median wall. In shape this Organ is generally speaking oval, but much drawn out and bluntly pointed posteriorly, while anteriorly it often ends quite abruptly. The length of its lumen is 1.2 mm., its total length being 1.4 mm. Its ventral edge is almost straight, the dorsal edge curving downwards posteriorly to meet the former. Its outline in transverse section varies considerably. Posteriorly, it is much flattened from side to side, its lateral wall being in parts slightly indented, though it can scarcely be called kidney-shaped (Figs. 5 and 6, J.O.). This to a certain extent is comparable with that shown for part of the Organ in Miniopterus,^ by Broom, by Klein in the Dog,' and in a much less degree with that shown by Symington and Smith, in Ornithorhynchus^ and Echidna,* and by Broom in marsupials generally. But, whereas in the former of these it is due more or less to an incurving of Jacobson's cartilage, in Notoryctes it is simply due to a thickening of the subepithelial 1 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1896, pi. xlvii., fig. 11. 2 Proc. Linn. Soc. N.S.W., vol. x. n.s., 1895, pi. 47, fig. 4. 3 Q.J.M.S., vol. xxii., p. 305. 4 Proc. Zool. Soc, 1891, p. 579. 5 Anat. Anz. XI. Band., 6, 1895, p. 162-3. 88 Proceedings of the Royal Society of Victoria. layers of the lateral wall, there being no inturning of the capsule. In the Rabbit' and ]\Iarsupials, however, there is a similar somewhat kidney-shape in the central portion of the Organ, due only to sul)epithelial and glandular thickening. This indentation, moreover, is not constant, as, occasionally, as above stated, while one side i-etains more or less of the concavity in its lateral wall, the Organ of the other side may be quite oval or even circular in transverse outline. In vertical diameter the Organ varies from .32 to .6 mm. Horizontal diameter, .12 to .28, very slightly less than in the Rabbit, and slightly under half of that of the Dog and Guinea- pig. Into the upper and lower sulci of the Organ there open a considerable number of ducts from the gland mass on either side of the nasal septun). Seven or eight of such ducts may at times be seen in one single longitudinal section opening into the upper or dorsal sulcus, and a lesser number into the ventral sulcus. These ducts, which are short, wide, and have darkly staining walls, lie at right angles to those from the same gland mass, which run forwards longitudinally, and more or less parallel, till they open into the vestibule close to the external orifice. The latter longitudinal ducts are usually fifteen to twenty in number on each side of the cartilaginous septum. Blood Vessels. Jacobson's Organ is well supplied with these (Figs. 5 and 6, f., a., ct.^. Alongside its lateral wall, there run an artery and two veins, the former curving round anteriorly, from dorsal to ventral surfaces, and between the front end of the Organ and its supporting cartilage in this region, while both laterally and ventrally in the median wall is to be found a more or less extensive plexus of blood vessels (Figs. 5, 6, c.t.). Nerves. In longitudinal sections, especially, there is to be noted a large branch of the olfactory nerve passing forward horizontally and entering into relation with the dorsal and mesial surfaces of the 1 Q.J.M.S., vol. xxi., p. 558. Anatomy of Notorycfes typhlops. 89 posterior part of the Organ, descending anteriorly to the mesial wall, as seen in transverse sections (Fig. 6, n.f.). Minute Structure of Jacohsoti' s Organ. For convenience of description we may take first the lateral wall, with the structures outside this, and then similarly the median wall. 1. — The Lateral Wall. The epithelium lining the Organ of Jacobson on this side (Fig. 6, l.w.) is .04 to .06 mm. thick, being slightly less than in the Dog, and the same as in the Guinea-pig and Rabbit. Tt con- sists of a columnar epithelium, similar to that lining the nasal cavity (which is .06 mm. thick), having here apparently two layers of cells ; (a) an outer columnar layer with long, strong cilia {c.f., Guinea-pig and Dog, and contrast the Rabbit), and oval nuclei. These are interspersed with goblet cells, which are numerous in parts of the lower half of the wall ; (/?) an inner layer with rounded nuclei. It will be seen that this differs from that of the Guinea-pig as described by Klein^ in that his middle layer of spindle-shaped cells is not visible here. Probably this is due to the fact that all the material at my disposal is spirit-hardened, and in such cases Klein has found great difficulty in distinguishing the spindle-shaped cells from those of the columnar layer. Next to this is a well-marked fibrous layer corresponding to the subepithelial layer of other forms, with blood vessels and gland alveoli. The cavernous tissue shown by Klein to be so well developed in this position in the Guinea-pig^ and Rabbit,'^ and by Broom in Phascolarctos,' and in Petauroides,* does not exist here in Notoryctes, the blood vessels of this side being limited to an artery (o), running longitudinally along the middle line of the tube, and one or two small veins. This is more like what we find in the ordinary Marsupials, which have a single hilar blood vessel. It may be seen in Macrosce- lides,"* and is much greater in extent than in the lateral wall of 1 Q.J.M.S., vol. xxi., p. 101-3. 2 Loc. cit. p. 563-4. 3 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1896, p. 613. 4 Loc. cit., p. 607. 6 Proc. Zool. Soc, 1902, vol. i., p. 226. 90 Proceedings of the Royal Society of Victoria. Miniopterus.^ The glandular development is here {m.g.), as in the Rabbit,- and, contrasted with Miniopterus^ and the Guinea-pig, most marked in the cartilaginous capsule, at the upper and outer part of the Organ, though unlike the Guinea-pig, where the glands are more numerous when the cartilage is alisent, in Notoryctes there seems to be no such invariable relation. There are also, as described above, numerous glands lying in the inferior septal ridge {i.s.r.) outside Jacobson's cartilage. In this respect, No- toryctes agrees with Didelphys murina,* Trichosurus,* and Dasy- urus maculatus,® while differing from the Phalangers generally, and from Perameles' and Dasyurus viverrinus.* At the same time, we find the general Diprotodont feature, characteristic also of Phascolomys,' in which numerous gland ducts open into the Organ from above. With regard to these glands around Jacob- son's Organ, it may be remarked that they appear to be regarded by Klein, as also those on the septum, as true serous glands in the Rabbit^'^ and Dog;'^ while Broom finds, in the septum, mucous glands in Miniopterus,^ and in various Marsupials also.^^ In Notoryctes, those in the mucous membrane of the septum and ridges covering the turbinal bones, appear to be true mucous glands, though those around the Organ of Jacobson and a small group on each side of the bottom of the nasal septum are appar- ently serous in character, and have smaller alveoli, more deeply staining nuclei, broader, deeply staining ducts, which all open into Jacobson's Organ. The gland ducts chiefly enter the tube at the upper and lower sulci, though occasionally they open through the lateral wall itself as previously found in the Rabbit'" and Sheep. Their number would account for the fact that the tube is always 1 Proc. Linn Soc. N.S.W., vol. x., n.s., 1895, p. 574. 2 Q.J.M.S., vol. xxi., pp. 563-4. 3 Loc. cit., p. 103. 4 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1836, p. 598. 5 Loc. cit., p. 607. 6 Loc. eit., p. 596. 7 Loc. cit., p. 602. 8 Loc. cit., p. 596. 9 Loc. cit., p. 613. 10 Q.J.M.S., vol. xxi., p. 564. 11 Loc. cit., vol. xxii., p. .306. 12 Loc. eit., vol. xi. n.s., 1896, p. 614; Trans. Roy. Soc. Edin., vol. xxxix., p. 233. Anatomy of Notoryctes typhlops. 91 full of secretion. From this, as from the large size of the Organ, we may perhaps infer that in Notoryctes the glandular function is relatively more important than the sensory one. Coming down from the side of the septum, and running longi- tudinally, are a small nuniVjer of scattered nerve fibres similar to those described by Klein in the Rabbit.^ The main features of the histology also agree closely with those described by Symington in Macropodidae.* 2. — Median Wall. The sensory epithelium lining this wall (Fig. 6, m.m.) extends also as described in the Guinea-pig by Klein^ in the anterior half, a short distance down the lateral wall of the superior sulcus, but ending at the angle of the inferior sulcus for the whole length. Its thickness varies from .08 to .1 mm., slightly greater than in Dog, and slightly less than the Guinea- pig. In the posterior part the sensory epithelium ends also at the angle of the superior sulcus. The boundary between the epithelium of the lateral wall and the sensory epithelium of the median wall is always very sharply marked otf. The sen.sory epithelium in Notoryctes resembles closely in its general structure that of the Guinea-pig,* Rabbit^ and Dog,* though the minute structure of the cells cannot be made out in these spirit specimens. The epithelial cells which bear short cilia appear much longer and thinner than those of the lateral wall, and have a striated border, probably due to the terminal rods of the cells in the lower layer. These epithelial cells have oval nuclei, which are disposed in three ill-defined layers similarly to the above mentioned forms. The sensory cells have large spherical nuclei more transparent and less deeply staining with haematoxylin, and with a well-marked nuclear membrane and network. They are arranged in one or two layers (as in the Dog), usually in one layer near the upper and lower sulci, and tw(t layers in the median part of the wall. In one or two places 1 Q.J.M S., vol. xxi., pp. 556, 564. 2 Jour. Anat. and Phys., vol. xxvi., p. 373. 3 Q.J.M.S., vol. xxi., p. 105-6. i Loc. cit., p. 564, etc. 5 Loc. cit., vol. xxii., p. 307-310. 92 Proceedings of the Royal Society of Victoria. in the length of this wall, the gland ducts pass through to open into the tube, the last part of their wall being lined by a continuation of the sensory epithelium. The space between the Organ of Jacobson and the cartilage of Jacobson, or the bone of the crista nasalis, is closely packed in its upper half with nerve fibres (Fig. 6, n.f.). These are much more numerous in the median and posterior portion of the wall, decreasing in quantity anteriorly. At the hinder end of the tube, a large bundle passes off to run in the septal mucous membrane until finally it joins the main olfactory trunk. I have been able to trace these fibres among the cells of the sensory layer, but not actually into the cells, where doubtless they do end. As the nerve fibres decrease in number their place is taken by glands. The cavernous tissue {c.t.) so conspicuous in Klein's figures of the lateral wall in the Guinea-pig^ and Rabbit, '^ and much more rudimentary in the median wall of the Dog,' is very abundant in the lower half of the median wall in Notoryctes. Here there are one or two arteries and several somewhat large veins forming a plexus, and supported by ordinary loose fibrous tissue. In the position of the nerves and veins in this median wall, we may compare this with Phasco- lomys.' In Notoryctes, as previously stated, the distinction between the medial and the lateral epithelium persists right to the posterior end of the Organ, as contrasted with the Rabbit, where only columnar epithelium is found at the posterior end of the Organ, and with Phascologale^ and with Macroscelides.^ Jacobson's duct, as heretofore described, is extremely short, and is lined by a continuation of the ordinary nasal epithelium similar to that of the lateral wall. This is to be contrasted with the ordinai-y marsupial, e.g., Dasyurus,' in which Jacobson's duct is lined with squamous epithelium. Stenson's duct, however, is lined by stratified pavement epithelium continuous with that lining the palate. The surface 1 Q.J. M.S., vol. xxi., pi. vii., fig. 5, pi. xvii., fig. 6. 2 Loc. cit., pi. XXX., fig. 5-8. 3 Loc. cit., vol. xxii., pi. xxvi., figs, 14, 15. 4 Proc. Linn. Soc. N.S.W., vol. xi., n.s., 1896, 614. 5 Proc. Linn. Soc. N.S,W., vol. xi., n.s., p. 594. 6 Proc. Zool. Soc, 1902, vol. i., p. 226. 7 Proc. Liun. Soc. N.S.W., vol. xi., n.s., 1896, p. 595. Anatomy of Notoryctes typhlops. 93 layers of the lining of the duct are strongly corneous, this diminishing, as in the Dog," as it enters the nasal furrow, to one-third of its thickness on the palate. The bottom of the furrow near Stenson's opening is similar to that of the duct itself. There are no glands opening through the wall into the canal of Stenson, as found in the Sheep (Balogh) and Man (Kolliker), but which Klein was unable to find in the Guinea- It is worthy of note that the stratified pavement epithelium lining the vestibule of the nose, which may be up to .1 mm. thick, has a very thick corneous layer which may be in itself .04- mm. thick, the epithelium covering the snout itself being up to .25 mm. in thickness, of which the corneous layer makes up .1 mm. Summary and Relations to other Forms. The chief points^ to be considered in discussing the i-elations of the Organ of Jacobson in Notoryctes to that of other forms are : (1) the direct or indirect connection of Jacobson's duct with the naso-palatine or Stenson's duct; (2) the presence of the outer bar of Jacobson's cartilage ; (3) the pre.sence or otherwise of a cartilaginous bar of support for the naso-palatine canal ; (4) presence or otherwise of the outer nasal floor cartilages behind the naso-palatine canal ; (5) the papillary cartilage of the prenasal cartilage ; (6) the arrangement of the blood vessels. I. — Jacobson's Duct, in Notoryctes, is seen clearly to open directly into the nasal furrow, from which in turn Stenson's duct leads down to the oral cavity. Here then we have the structure regarded by Broom as typical of the Rodents,^ and also found by him in one Diprotodont form Aepyprymnus, and in Dasypus among the Edentata. In reference to this feature in Aepyprymnus*, Broom considers it to be only a slight difference in the relative position of these openings, " due to the lengthening of the front of the snout in connection with the well-developed front iiicisors." But it cannot be so caused here in Notoryctes, 1 Q.J.M.S., vol. xxii., p. 3U1. 2 Q.J. M.S., vol. xxi., p. 229. 3 c.f.. Broom : Proc. Zool. Soc. of London, 1902, vol. i., pt. ii., p. 226. 4 Proc. Linn. Soc. N.S.W., vol. x., n.s., 1895, p. 572. ."i hoc. dt., vol. xi., 1896, p. 619, and Trans. Roy. Soc. Edin., vol. xxxlx., p. 241. 94 Proceedings of the Royal Society of Victoria. and one is led to think that there is more meaning in its presence tliere also, than that of mere parallel development in two animals possessing a rodent type of dentition, even if, as stated by Broom,' we are to regard Aepyprymnus as " approximating to a rodent type of dentition." II. — We have seen that in Notoryctes, for a part of the length of its lateral wall, it has a more or less convex character, so constricting the lumen of the Organ, this being due, not to a well marked cartilaginous support as in Ornithorhynchus, or to an incurving of the edge of the cartilage as in Echidna, or even in Miniopterus or the Dog, but to a thickening of the subepithelial and glandular layers of the lateral wall, forming a "gland fold," as in the Rabbit and Guinea-pig, and in its near allies, the Marsupials. Also we find in this lateral wall the outer bar of Jacobson's cartilage as a ridge process, which is undoubtedly the rudimentary homologue of the turbinal found in the more highly organised structure of Ornithorhynchus, and to a less extent in Echidna. In a more or less developed form this outer bar is found in all Marsupials. In the degree of development found here, Notoiyctes is most closely allied with the Phalangers, especially Petaurus in which it is more developed than in the Polyprotodont Dasyure, and with the Macropods to a less extent, and witli Dasypiis and the Rodentia among the Eutheria. III. — The cartilaginous support for the naso-palatine canal in marsupials is never more than rudimentary ; and even so, as in Perameles, Trichosurus, Phascolarctus, Macropus, Phascolomys, and Petaurus, it is always on the inner side and not on the outer anterior side of the canal, as in Notoryctes ; and also in the Rodents, in which, however, it is much larger than that in Notoryctes, and in Miniopterus among the Cheiroptera. IV. — In Notoryctes, as slightly different from other Marsupials, and Edentates, there is a very fragmentary continuation of the hinder edge of the outer nasal floor cartilages for a short distance behind the opening of the naso-palatine canal. In a degree, this may indicate a leading-on to the Rodent type, in which the cartilage persists behind the plane of the naso-palatine canal. It must be remembered here, also, that in the attachment of the 1 Trans. Roy. Soc. Edin., vol. xxxix., p. 241. AnatoTTiy of Notoryctes typhloiJS. 95 nasal floor cartilages anteriorly to the septum, we find a similar condition in Ornithorhynchus and Echidna only. V. — Though not developed to nearly the same extent that it is in Ornithorhynchus, or Echidna, or even in Miniopterus, there is, I think, undoubtedly a j)renasal cartilage present in Notoryctes. It certainly cannot be called a prolongation anterior to the carti- laginous nasal septum, and since that in Notoryctes, extends right forwards to the end of the snout, it could not find room there. But it does split ofi" from the ventral edge of the anterior part of tliis septum, and its hyaline nodule in position exactly corres- ponds to that found in the foetal calf. In part it also corres- ponds to the well-developed prenasal found in Miniopterus, since in each there is a central more or less fibrous ridge between the palatine processes of the premaxillae, giving off in Miniopterus, and to a certain degree in Notoryctes, a lateral sheet to support the papilla between Stenson's ducts, this latter somewhat resem- bling Marsupials, though in them the centi-al ridge is absent. In Notoryctes, the prenasal is less developed than in Miniopterus, though exactly similar in relations to the surrounding cartilages and bones, because in the former the preraaxillaries come together and fuse further back, and so shut out the possibility of the exist- ence of any prenasal there, whereas in Miniopterus they do not meet in the middle line. One may here remembei', also, that the nodule of hyaline cartilage, described by Klein in the Guinea-pig, supporting the papilla, is, as stated by Broom, probably to be regarded as a remnant of the lateral sheet of the prenasal cartilage. VI. — Typically, in the Marsupialia there is to be found a single large vessel running along the outer face of the Organ. In Notoryctes we find two or three distinct vessels in this position, and a well-marked plexus in the median wall. In Polyproto- donts generally, this plexus is rudimentary, and in the lateral wall, in Diprotodonts, it is generally well marked, as also in the Edentates, while in the Rodents we often find a very large vascular plexus in this wall. Probably, as observed by Broom, this feature is not of much importance in classification, since in such closely allied forms as the Mouse and Guinea-pig, we find con- siderable differences. Similarly with the glands, though Broom has considered that the large vascular plexus, and the numerous 96 Proceedings of tJte Royal Society of Victoria. glands present in Rodentia, point to an affinity with the lower Mammals. Further, he finds a great glandular development to be typical of large forms, e.g., Lepus and Trichosurus. Here we have such in a small form. Here, doubtless, the numerous glands opening into the lumen of Jacobson's Organ are associated with the great amount of glandular material covering the septum, and the turbinal ridges, as is also the remarkable development of glands in connection with the degenerate eye : though I do not consider that, in the case of Jacobson's Organ, this great secre- tory power is necessarily developed at the expense of the sensory function, as in the eye — since we find in Jacobson's Organ here consistently with Broom's generalisation that the Organ is more highly developed in small forms than in large — it is in Notoryctes well developed, occupying fully two-thirds, and in parts the whole, of the cai'tilaginous trough in which it lies. Conclusion. It would seem then from the evidence of Jacobson's Organ, that we are justified in claiming for the Polyprotodont Notoryctes, that, while it still has traces of a Monotreme relationship, it shows a close affinity with the Diprotodonts by way of Aepyprymnus and Petauruis, and also, though at a much greater distance, with the Edentates and Rodents. It thus adds its measure of confirmation to the position given by Broom, as doubtful as yet, to the Rodentia in his classification of the Mammalian groups, in which he classes the Edentates and Rodents under one group, the Archaeorhinata. It, pari passu with this, adds its testimony to that of the muscular system, which has been held by Professor Wilson,^ to show " enduring evidences of a real, if distant, morphological kinship " with that of the Edentates. Part II. Blood Vascular System. This system, while not showing so far many special points having a general significance, has still a number of interesting conditions which are well worthy of record, in addition to the 1 Trans. Roy. Soc. S. Au?., 1894, p. 5. Anafoviy of Notorycfes typhlops. 97 normal coiiclitions present. The study of the blood vessels, with the material at present obtainal)le, is not an easy matter, owing partly to the very brittle and absolutely bleached, and often quite transparent state of the vessels, and also to the great quantity of adipose tissue surrounding them, with a considerable admixture of strong fibres, which to the naked eye are often much more like ordinary blood vessels than are those vessels themselves. Recourse has frequently to be made therefore to the compound microscope and staining fluids for certainty of recognition. Especially is this so in the pectoral, abdominal, and pelvic regions. The following details involve observations made during a careful dissection of five individuals, aided by microscopic sections of one or two parts, such as the limbs. 77/*? Heart. The heart, which is normal in position, is somewhat more pointed than is often the case, the apex being well directed towards the left side, and separated dorsally from the diaphragm by a small lobe of the right lung, as in marsupials generally, its pericardium, however, being distinctly connected ventrally with the diaphragm, a condition not usual in marsupials. So far as can be seen there is no fossa ovalis on the auricular septum. In the left ventricle the mitral valve has two well-marked papillary muscles holding its chordae tendineae, one on the septum, the other on the outer wall, while the right auriculo-ventricular valve has three muscles corresponding to its three flaps. Tl>e right ventricle takes no share in the formation of the apex. Pul»ionary Circulation. The main pulmonary artery is a short thick vessel arising from the right ventricle, and leaving the heart externally just behind the arch of the aorta. It divides almost immediately, and at a point directly ventral to the trachea, and anterior to its division, into the right and left pulmonary artery. The right branch is somewhat shorter and wider than the left, each of the branches lying ventral and somewhat anterior to the bronchus of its own side. Each artery divides at its entrance to the root of the lung into two main branches, the larger of which passes downwards 98 Proceedings of the Royal Society of Victoria. to the lower part of the lung, alongside the main branch of the pulmonary vein. The pulmonaiy veins are two large vessels, each of which is formed, as it leaves the lung of its own side, of two, or some- times three, main vessels. The left pulmonary vein would appear to be both longer and wider, as also more sloping, than the right. The two unite and form a median trunk, as in Marsupials generally, similar in thickness to the corresponding trunk of the pulmonary artery, but at least twice as long as the latter. Each pulmonary vein runs ventral and also posterior to the bronchus of its own side. The division of the trachea into the bronchi occurs dorsal to the anterior half of this main pulmonary venous trunk. It then opens into the i-ight auricle by a wide aperture, behind the emergence of the pulmonary arterial trunk, and in front and slightly to the right of the entrance of the left anterior vena cava into the right auricle. This is the usual arrangement of these parts in Marsupials. Systemic Arteries. The aorta emerges from the base of the heart at about the same level vertically, or slightly in front of the pulmonary artery, curving towards the front and left, round the trachea, and then backwards dorsally to the bronchus and root of the left lung. From the beginning of the arch, as usual, the coronary vessels are given off, one of which only can sometimes be seen with the unaided eye. The relative positions of the origins of the carotid and subclavian arteries vary somewhat in different individuals. The two types are : (1) The two carotid arteries, left and right, arise as a common trunk ^-inch in length, from the root of which opens the right subelavian artery, the left subclavian leaving the arch considerably to the left end of the transverse part of this arch. This corresponds to the condition found in the majority of Marsupials and in Choeropus in particular.-^ (2) In other speci- mens again, and, so far as my material shows, most frequently, the right carotid and subclavian arteries arise as a common inno- minate trunk similar to that of many higher forms, including Man. The left carotid artery arises close to the base of this 1 Parsons : Jour. Linn. Soc. Lond., Zoology, vol. xxix.,No. 188, Oct. 1903, p. 64. Anatomy of Notoryctes typhlops. 99 innominate vessel, the left subclavian having its origin some little distance to the left of the left carotid artery, and not close beside it, as in Man. In this, Notoryctes resembles the broad- chested Marsupials, such as the Wombat and Koala.^ Consequent on these variations the lengths of these vessels vary also. In relation to the nerves, the carotid artery lies ventral to the recur- rent laryngeal, and pneumogastric nerves, crossing them obliquely as it runs outwards towards its anterior end. The sympathetic nerve appears to lie quite to the other side of the coiiniion carotid artery on each side. Where the two common carotids and the right subclavian artery are united at their origin from the aorta the pneumogastric also lies to the outer side of each common carotid, since the angle caused l)y this vessel in its course for- wards is then considerably greater than where the vessels of the right side only are united to form an innominate vessel. The common carotid gives off no branches, but divides anteriorly into the external and internal carotids. The e.xternal carotid lies at first slightly below and distinctly nearer the median line than the internal carotid. It soon gives off the superior thyroid artery, which runs straight forwards and inwards to the thyroid gland. The ascending pharyngeal artery appears sometimes to be given off from the internal carotid just anterior to the bifurcation of of the common carotid, instead of being associated with the external carotid, as in higher forms. A little in front of the superior thyroid, the lingual artery is given off, running above the digastric and stylo-hyoid muscles, and continues under the mylo-hyoid muscles, giving off a branch to them, and then supplying the tongue and contiguous parts. Just where the lingual artery is given off, the external carotid turns outwards, curving round behind the masseter muscle. On its posterior side, as it curves round the articulation of the jaw, the external carotid gives off" the occipital and posterior auricular arteries, while from its anterior side is given off the facial artery, the main vessel then breaking up into temporal and internal maxillary arteries. The four last mentioned arteries leave the main trunk very close together, the occipital arising about half way between these and the origin of the lingual artery. It will thus be seen that the 1 Owen : Anatomy of Vertebrates, vol. iii., p. 539. 7a 100 Proceedings of the Royal Society of Victoria. lingual and facial arteries are much farther apart than is very often the case. Also, the facial here arises quite independently of the temporal artery as contrasted with Choeropus.^ The internal carotid, as previously stated, gives off immediately beyond its origin from the common carotid, the ascending phalangeal artery, a condition to be contrasted with the normal origin of this artery from the external carotid trunk. This internal division of the carotid trunk runs down deeply, external at first to the external branch until it lies close along- side the pneumogastric nerve and the superior cervical ganglion, and passes forwards between the muscles to enter the skull. The varying relations of the subclavian artery to the main aorta have already been described. The vertebral artery appears to be similar in position to that of other marsupials, but is generally very small, and often invisible. The inferior thyroid artery and its branches are, compared with their usual propor- tions, very slender, especially when contrasted with the internal mammary artery, which often approaches the main subclavian artery in size. The deep cervical and superior intercostal arteries leave the subclavian trunk separately, the latter being- proximal to the former. The long thoracic, posterior scapular and subscapular, are all normal in po.sition but of considerable size. The brachial artery divides early into ulnar and radial branches : the relative position of these to each other, and to the nerves and muscles of the forearm, conform in general to the usual mannnalian type, as do also the branches and palmar arch of the large median ulnar artery, so far as they could be made out either by dissection or by sections. Here, as contrasted with the majority of marsupials, the ulnar branches pass over the condyle of the humerus instead of piercing it. The thoracic aorta passes round dorsally to the left bronchus, find then posterioi-ly, in close contact with dorsal wall of the thorax, to which it gives oif a few very small vessels, then piercing the diaphragm to enter the abdomen. Abdominal aorta.— This gives off the coeliac artery, which is long and divides into well-marked gastric, splenic and 1 Parsons : Jour. Linn. Soc. Lend., Zool., vol. xxix., No. 188, 1903, p. 64. Anatomy of Notoryctes tyi^hlcps. 101 liepatic branches, and somewhat lower, a much larger vessel the superior mesenteric artery, which gives off a distinct inferior mesenteric artery to the lower parts of the intestine. Below these are the renal arteries (Fig. 8, r.a.)^ the right being small and short, the left long and broad. The spermatic arteries {g.a.) are very small. Near and posterior to the renal arteries, as so often in Marsupials, the aorta lies quite dorsal to the posterior vena cava, by which it is coujpletely hidden, until some distance below its bifurcation, when the external iliac arteries come to lie, still somewhat dorsaily but more to the outer side of the external iliac veins. From the following description it will be seen that there is a great difference be- tween this region in Notorcytes and in the Marsupial type, in which the abdominal aorta, after giving off the external iliac arteries, continues back, giving off the two internal iliac arteries, the small continuation then forming the median sacral artery, e.x'-i in the Kangaroo' and Ohoeropus.'^ In Notoryctes we find that the aorta bifurcates to form the common iliac arteries {c.c), and in front of this bifurcation, from the dorsal wall of the aorta we may get a very small median sacral artery (^m.s.), often only to be found by removing the neighbouring tissues, staining and examining them vinder the compound microscope. At other times I have found two, or in one case, three vessels, just visible to the naked eye, arising on either side, posteriorly, of the bifurcation, which from their distribution must represent the median sacral artery. At other times I was unable to detect any median sacral artery whatever. On its outer side each common iliac artery gives off what correspond in their distribu- tion to the ilio-lumbar arteries (/./.), and still further down there arises the circumflex (.c.) ; also the premaxillary {/>.d.) and nasal {/i.b.) bones. The naso-lachrymal duct (n./.d.) is seen pa.ssing through its canal immediately posterior to the union of the alinasals and the cartilages of the nasal floor {n.J.c), to open into the ventral nasal furrow anteriorly. Zeiss A,* oc. 2. Figure 2. Ventral median portion of trsnsverse vertical section through head, a considerable distance behind Fig. 5, and just in front of Jacobson's Organ, to show swelling in Jacobson's cartilage (_/.^.) which has, on the right side, quite lost its anterior connection with the nasal septum (n.s.c), and on the left is just losing it. The naso-lachrymal duct is seen {n.l.d.) to lie in its groove in the alveolar process of the bone outside the outer nasal floor carti- lages {n.J.c). The ill-defined papillary cartilage {f.p.c.) is seen ventral to the palatine processes of the premaxillary bone {p.p. p.). Zeiss A,* oc. 4. Figure 3. Complied from two consecutive transverse vertical sections through Jacob.son's Organ {/.O.) at the point where its duct Pi-or. R.S. Virtoria, 1904. Plate Vl. Proc. R.S Vlrtnrlii. 1904. Plate VII. Fi&:l, Proc. R.S. Victoria, 1904. Plate VIII. iir Pro,: E.8. VirtMria, 1904. Plate IX. Fi(J. S Anatomy of Notoryctes typhlops. 109 {/■d.) opens into the ventral nasal furrow {v.ii.f.), in the same plane as the ventral opening of the naso-palatine canal {ii.p.d.) into the buccal cavity. The outer bar of Jacobson's cartilage {o./.c.) is here seen separated from the vertical part of the enclosing cartilage, and anterior to its union with the ventral outer edge of Jacobson's cartilage, arching in this section over Jacobson's duct. The gi'eat development of mucous glands is also evident hei'e. Zeiss A,* oc. 2. Figure 4. Portion of transverse vertical section, three posterior to Fig. 7, showing connection of outer bar of Jacobson's cartilage {o./.c.) with ventral outer edge of the now crescentic Jacobson's cartil- age {/.c). The opening of the naso-palatine canal {71. p. d.) from the ventral nasal furrow {v.n.f.) is clearly shown, as also the rudimentary cartilages of the nasal floor {ii.f.c). The naso- lachrymal duct is seen lying in a definite canal in tlie alveolar bone, below tlie maxillo-turbinal process. Zeiss A,* oc. 4. FlGURK 5. Section similar to last, but a little posterior to it, showing Jacobson's cartilage {o./.c.) diminishing on the left, and quite re- placed by a bony shelf {p-P.s.) from the palatine processes of the jiremaxillary bones {p.p.p.) on the right. Zeiss A,* oc. 4. Figure 6. Ti-ansverse vertical section through Jacobson's Organ {/O.), showing more minutely its structure and relations. The differ- ence between the lateral {Lw.) and medial walls {m.zv.) may be noted, as also the artery [a) and vein (v) in the lateral wall, and the cavernous tissue (c.t. and v.) in the ventral part of the medial wall. The bundle of nerve fibres is seen descending, to be dis- tributed to the medial wall from the main Jacobson's branch {n.f.) of the olfactory nerve. The mucous glands {ni.g.) of the mucous membrane may be contrasted, in appearance and struc- ture, with those Ijelow, which have much smaller alveoli. The bony slielf {p-p.s.) from the palatine processes is clearly seen. Zeiss A., oc. 4. 110 Proceedings of the Royal Society of Victoria. Figure 7. Ventral view, showing kidneys, testis, and bladder, and veins associated vi'ith them. Approximately x 3^. Figure 8. Ventral view of abdominal aorta, showing the vessels of the pelvic region. Approximately x 4. APPENDIX. June 18th, 1904. A more complete paper by Professor McClure on the Anatomy of the venous system of Didelphys marsupialis^ (called in his previous paper here referred to, D. virginiana) has come under my notice, and it seems desirable to make some reference to it. (1.) McClure adds yet another to the list of Marsupials in vv'hich there is a common internal iliac artery — in Petrogale^ (sp-'?) as contrasted with Notoryctes (see Fig. 8). In Petrogale also, as in Didelphys, there is a well marked median sacral artery such as is not found in Notoryctes. Further, in Didelphys these arteries appear to lie in general internal to the veins, sometimes dorsal, and sometimes ventral to them, whereas in Notoryctes the arteries lie external and generally dorsal, being rarely, if ever, quite ventral to the veins. With these exceptions, the division of the abdominal aorta in Didelphys into two common iliac arteries, and of these into external and internal branches, is very similar to the condition found in the Notoryctes, and quite unlike that found in Petrogale among other Marsupials. (2.) It will be found that though some of the variations in tlie relations of the iliac veins in Didelphys are very much like those of Notoryctes (notably PI. II., Fig. 8, being one variation of McClure's Type II.), yet even here Notoryctes differs from Didelphys in the most posterior union of the common iliacs to form the posterior veua cava, and also in the position of the 1 Amer. Jour, of Aiiat., vol. ii., No. 3, 1903, p. 338, fig. vi. Anatoviy of Notoryctes typhlops. 1 1 1 arteries. So that the condition of these blood vessels in Notoryctes does not by any means fit into any one of the types of variation found in Didelphys. Indeed, so far as one isjustitied in relying on an individual figure or dissection, the condition of the iliac veins, and the relation of the external iliac artery to the external iliac vein, shown by McClure in Petrogale^, is much more like Notoryctes. It is evident that more detailed knowledge is needed of the condition of these blood vessels in Marsupials generally before they can be relied upon as a final test of aflinity between groups. 1 Amer. Jour, of Anat., vol. ii., No. 3, 1903, p. 338, fig. vi. Art. III. — TJte Relations of the Granitic and Lower Palaeozoic Rocks near Dandenong. By IAN M. SUTHERLAND (Communicated by Professor J. W. G-regory, D.Sc, F.E..S.). (With Plate X.). [Kead 12th May, 1904.] I.— Object. The granites of Victoria are of two ages ; one granite is pre- silurian, and the other was intruded in the earlier part of the devonian age. Mr. A. W. Howitt writes •} " Tlius, leaving out of the question those rocks which are clearly felsites, it becomes evident that there are in North Gippsland two distinct classes of granites. The older are truly granitic in character, and frequently hornblendic as well as micaceous ; the younger approach nearer to the felsites, and, so far as I am aware, are not only poor in mica, but also quite without hornblende. The older granites may approximately be placed at the close of the silurian, and the younger granites in the earlier part of the devonian age. There seem therefore to be grounds for the statement that, so far as our present knowledge extends, the devonian granitic rocks of North Gippsland have a peculiar character wherever met with ; Init in this it is necessary to guard strongly against the supposition that no true granites may have been formed in that age." In memoirs of the Geological Survey of Victoria, in an appendix to the Report on the Chiltern Goldfields, by Stanley B. Hunter, page 42, Professor Gregory writes : "Mr. Howitt long since suggested that the granitic rocks of Victoria belong to two distinct groups. Those of one group were intrusive in devonian times. The earlier group was pre- silurian. It has been the custom to regard the great majority of the Victorian granitic rocks as belonging to the devonian group." 1 Report of Proj^ress of the Geological Survey of Victoria, 1877. Notes on the Geology of Part of the Mitchell River Division of the Mining District of Gippsland, p. 121. Granitic and Palaeozoic Mocks, Dandenong. 113 The age of the granite at Walhalla has an important bearing on the geology of the goldfield of that district, but no fully satisfactory evidence of its age has hitherto been got near there. The granite is marked on the Geological Sketch Map of Victoria as extending almost continuously from Dandenong to Mount Baw Baw and Walhalla, and therefore its age at these three places is probably the same. As the two groups of granites are not distinguished in the geological maps of Victoria, at the suggestion of Professor Gregory, I have examined the granite near Dandenong to try to determine its age relative to the lower palaeozoic beds.^ II. — Topography. Dandenong is eighteen miles south-east of Melbourne, on the creek of the same name, which flows from the swampy land to the north-east of the Dandenong ranges. The township is about 69 feet above sea level, on slightly undulating country, to the north-east of which rise the " Dandenong Ranges "; but as these hills are isolated or connected by very low saddles, a better name would be the Dandenong Hills. Near Dandenong there are three types of rocks : — 1. Granite, forming the picturesque foot hills of the pro- posed National Park, in the old Police Paddocks, and the hills to the east and north-east. 2. Dacites, similar to those of Mount Dandenong, occurring near Ferntree Gully. 3. Lower palaeozoic rocks, on the flanks of the granite, and forming the low, undulating country towards Oakleigh. III. — Literature. There is not much literature on the geology of Dandenong district. The earliest is a report by the late A. R. C. Selwyn on a geological map of the country between Port Phillip Bay and 1 In the absence of fossil evidence there is no certainty whether the rocks be ordovician or Silurian, but the discovery, by Mr. Ferguson, of ordovician fossils in the Mornington peninsula suggests the possibility that the lower palaeozoic rocks in contact with the granites at Dandenong are also ordovician. In that case the Dandenong-Baw Baw massif may be pre-silurian though post-ordoviciaii. 114 Proceedings of the Roijal Society of Victoria. Westeniport.^ The Dandenong Creek forms part of the eastern boundary of Selwyn's map, and as all the granite, except one very small outcrop, is on the east side of the creek, the only mention of the Dandenong plutonie rocks is : " These two rocks" (feldspar-porphyry and syenite) "occur as narrow dykes cutting through and upheaving the older palaeozoic rocks, the former being on the south banks of the Yarra, and to the north of Mel- bourne, and the latter near Dandenong." The land to the south of Dandenong is described in the map as palaeozoic sandstones, shales, clay-slates, etc. A dyke of "sienite" is shown crossing the Dandenong creek, near Dandenong township. In the Report of the Geological Surveyor on the Geological Structure of Victoria, 1855-56, Sec. 3, page 17, Selwyn, in his description of the plutonie rocks south of the Yarra and east of the Dandenong Creek, writes: "They have upheaved and metamorphosed the palaeozoic strata on their flanks, and have therefore been intruded since the deposition of the latter. Whether the granite and porphyries are of different periods, or only accidental modifications in mineral character of the same mass is uncertain. They often appear to pass into each other, but small isolated patches of the porphyry, as well as branches from the mass, are found penetrating the granite, which is not found similarly intruding in the porphyry ; and we might there- fore imagine the porphyry to have been erupted at a period subsequent to the formation of the granite. " The granite near Dandenong is not descril)ed particularly in " Geology and Physical Geography of Victoria," by Reginald A. F. Murray. On page 27 he writes : " Among the areas represented as ' trap ' on the Geological Sketch Map, the rocks in three, namely, those of Mount Macedon, the Dandenong Ranges and Mount Juliet, besides others of minor extent, appear to be intimately associated with the ordinary granites, though the true relations of the rocks have not yet been properly investigated." On page 28, Mr. Murray mentions that specimens of "ternary granite" and " syenitic prophyry " from near Dandenong, and of " micaceous felspar trap," "felspar porphyry," and "syenitic felspar porphyry " from the Dandenong Ranges, are described in Mr. 1 Victoria. Votes and Proceedings, 1854-55, vol. i., pt. ii., p. 976. Granitic and Palaeozoic Rocks, Dandenong. 115 Selwyn's catalogue. In a paper published in the Proceedings of the Royal Society of Victoria, vol. xiv., pt. ii., page 211, Professor Gregory writes : " Mr. Ferguson has stated that there is a gradual change from the ' granites ' to the Dandenong ' traps ' ; but I have failed to find evidence of this, and Mr. T. S. Hart, who examined the sections on the Gembi'ook railway, tells me that wherever tlie two rocks could be seen together they were both greatly decomposed. He says there was no sign of a passage between the two rocks. This evidence is consistent with tlie view that the diorites and the dacites belong to different dates, and had independent origins." In a footnote Professor Gregory adds : " Since the paper was read I have examined the sections in question, and agree with Mr. Hart's conclusions." In the same paper (page 201) are given analyses of two porphyrites and a granodiorite from two miles north of Dandenong township, and a Dandenong dacite ; the analyses were made by the then Government Metallurgist, Mr. H. C. Jenkins, A.R.S.M. IV. — Geological Structure. Near Oakleigh the lower palaeozoic rock is a soft, yellow mud- stone, containing no fossils, and having a dip of about 18° W. The beds continue to the east without any change, except that of dip, as far as the Stud Road running north from Dandenong. On the west of this road the dip is constant for about three miles, and is 35" E.; it does not correspond to the surface formation, which seems to be due to erosion. About four miles north of Dande- nong the Lysterfield Road, running east and west, crosses the Stud Road. On the south side of tliis crossing, for about a mile along the Stud Road, the lower palaeozoic rocks are exposed by road cut- tings. The rock has the san)e appearance as that near Oakleigh, and the dip varies from 26° to 36° in directions between N. and and E.; the dip of the beds in these cuttings, and in all the other places where it can be seen, is nearly always towards the granite area and never away from it. The dip is very variable, and about a mile south of the Lysterfield Road there are some faults exposed by a shallow road cutting. On the east side of the Stud Road, on the Lysterfield Road, are a few cuttings showing lower palaeo- 8a 116 Proceedings of the Royal Society of Victoria. zoic rock. Near the Stud Road the rock is very much broken, and the beddin^ is indistinguishable, but the rock is not otherwise altered. The next cutting is about two miles to the east of the Stud Road, where the Lysterfield Road crosses the foot of a hill ; the rock shown has lost all sign of sti^atification, is harder than the unaltered rock, and is coloured red with iron. At the sum- mit of the hill small pieces of indurated I'ock are scattered about, some of them showing traces of stratification. A series of hills of similar formation lies to the south-west of this hill, towards the Lower Reservoir. In a quarry for road-making, near the foot of one of the hills, the altered rock is very hard, and has no apparent stratification. In one part of the face of the quarry there is some very decomposed rock, containing a good deal of mica- it underlies the altered I'ock, and may be a granitic dyke. The surface of some of the altered rock is encrusted with secondary mica. To tlie east and south-east of these hills there is a great number of smaller hills showing granite bosses near their sum- mits, but I could find no other rock except at two places ; tlie first is about half-a-mile south of the Lysterfield Road, and is marked (12) on the sketch map. At this point, near the foot of a hill, there is an outcrop of intensely altered stratified rock, containing white mica, and bedded vertically ; it has a strike N.N.E., or at right angles to the slope of the hill. The outcrop of granite begins about ten yards higher up the hill, but the actual contact is hidden by soil. The second place where other rock besides granite is exposed is marked (11) on the sketch map. A hill, showing large outcrops of granite, is crossed by a dyke about six feet wide, and exposed for about 100 yards ; the granite can be seen on both sides, but the line of contact is hidden by soil. In structure the dyke rock is much finer grained than the granite, and was therefore probably formed under a smaller pressure and later than the granite. Further south, on Bald Hill, loose pieces of rock, similar in appearance to that in the dyke, are found above the granite in a cutting, but no dyke can be seen. Granitic and Palaeozoic Mocks, Dandenong. 117 v.— Conclusions. As the granite is approached, the alteration in the ordovician or Silurian rocks is so marked that there can be no doubt that the granite is post-ordovician. Most of the stratified rock near Dandenong is hidden by a thick layer of loam and clay, so that the bedding can only be seen in a few road-cuttings and quarries. As stated above, the lower palaeozoic beds wherever exposed are found in almost all cases to dip away from the granite area. This formation seems to be common in Victoria, and Mr. R. A. F. Murray in his Geology and Physical Geography of Victoria, page 24, writes : " Another marked feature is that the granite intru- sions do not appear to be connected with the folding process to which the silurian rocks have been subjected, and to which is due the normal high rate of inclination of their layers. That process would appear to have taken place prior to the invasion of the sedimentary strata by igneous masses, as we find in many cases that the strike of silurian strata abuts directly on the granite, and in others that the dip of the strata is against, instead of with, the surface slope of the granite. Evidences of the intrusive character of the granite to a certain extent are, how- ever, visible in many places, in the locally contorted and crumpled state of the silurian strata, near their contact with the former." Mr. Murray then goes on to infer, from the description given by A. R. C, Selwyn, of the country east of the Snowy River, that much of the granite there was formed by the fusion and recrystallization of the silurian rocks. The granite near Dandenong does not seem to have been formed in this way as there is no intermediate rock between the granite and altered stratified rock. There is no evidence to show that the dacite further north, near Ferntree Gully, is contemporaneous with the granite. Since dykes have been found connected with the granite, but none with the dacite, the latter is probably the younger. Therefore the granite was formed after the ordovician period, but before the dacite. 118 Proceedings of the Royal Society of Victoria. Appendix. Note on the Microscopic Structure of Some Rocks from Dandenong, By professor J. W. GREGORY, D.Sc, F.R.S. In connection with Mr. Sutherland's paper on the lower palaeozoic and granite rocks of Dandenong, I have given a few of the rocks a microscopic examination. One of the lower palaeozoic rocks from allotment 61, Narre Warren, collected by the contact, proves, on microscopic examination, to have been altered into a very fine grained biotite hornstone ; it closely resembles some of the ordinary typical rocks formed by contact metamorphism around our granitic masses. The ordinary granitic rock of the district is connected with a series of dykes, the examination of which was of interest, owing to the possibility of some of them having been connected with the Dandenong dacites. The dykes examined, however, liave no connection with that series, and may be all derived from the much older grano-diorite massif. The dykes, of which the best occur in the Police Paddock and some adjacent allotments, belong to two groups. The first is a diorite-porphyry (No. 11), which is composed of phenocrysts of pale, partially leached, hornblende, and of andesine in a coarsely granular holocrystalline base. The second series of dykes is better described as quartz-biotite-porphyrite. Biotite is abund- ant, but has now been mainly altered into chlorite, the larger crystals containing granules of epidote, surrounded by the green chlorite. There are abundant corroded and embayed phenocrysts of quartz, and also of plagioclase. Tliese phenocrysts are widely scattered in a very fine-grained felsitic base, which was, no doubt, originally glassy. In some cases the felspars have undergone considerable decomposition, and the dull, dusty crystals, under polarised light, are lightened up by the bright granules of zoisite. The following analyses of these rocks may be conveniently repeated^: — 1 Gregory, J. W. : The Geology of Mount Macedon Victoria. Proc. Roy. Soc. Vict., vol. xiv., U.S., 1902, p. 201. Proc. R.S. Victoria, 1904. Plate X. 0 S E S ' ( •>(/ f ' f I I f ( i I I:; Ml ' / > \, '<^