24 J. P. HILL. 



who decline to accept Caldwell's interpi-etation to put forward 

 an alternative one, since I am unable to do so. 



The shell-membrane of Dasyurus (PL 1, figs. 8-11; PL 2, 

 figs. 17, 18, s.m.) is a transparent, perfectly homogeneous 

 layer, highly refractive in character and of a faiut yellowish 

 tint. When fully formed it possesses firm, resistant properties, 

 recalling those of chitiu, and is doubtless composed of a 

 keratin base. It is distinguishable at once from the albumen 

 by its 02:)tical characters and staining reactions, so that there 

 is not the slightest justification for the supposition that it 

 may represent simply the specially differentiated outermost 

 portion of that layer. In ova which have just passed into the 

 uterus (fig. 10) the shell-membrane is extremely delicate, its 

 thickness being only about 'OOIG mm., but even before cleavage 

 begins it has increased to "002 mm. (fig. 12) ; in the 2-celled 

 stage (fig. 18) it has reached "005 mm., in the 4-celled stage 

 (fig. 22) -0072 mm., whilst in tlie 16-celled stage (figs. 24-26) 

 it has practically attained its maximum thickness, viz., •0075— 

 •008 mm. Caldwell's measurements in the case of Phasco- 

 larctus agree closely with the above (shell of un segmented 

 ovum from the uterus, "0015 mm. thick, that of the '3 mm. 

 "ovum," "01 mm.). Its presence rendei-s the thorough 

 penetration of ova and early blastocysts with paraffin a 

 capricious and frequently troublesome operation, and its 

 resistant shell-like nature becomes only too obvious in the 

 process of section-cutting, since it cracks with the utmost 

 readiness (cf . PL 3, figs. 32, 37). 



The occurrence of a shell-membrane round the Marsupial 

 C)vum is a feature of considerable phyletic significance, as I 

 need hardly point out. It shows us that the ancestors of the 

 jMetathei'ia must liave been oviparous, or must themselves 

 have come fiom an oviparous stock, which there is no valid 

 reason for supi)osing was other than Prototherian in its 

 characters. It also renders untenable the views of Hubrecht to 

 the effect that tlie Metatheria are the descendants of Eutheria, 

 whilst the Eutheria themselves have been directly derived 

 from some presumed vivi]nir()ns group of hypothetical Pi^ote- 



