1909.] OLFACTORY ORGAN OF TELEOSTEAN FISHES. 659 



conical tube {Solea, Cycloi^terus, Trachinus), or the thin back- 

 wardly directed anterior lip of a slit-like orifice (Mugil, SphT/rcena, 

 Scomber, Bovichthys), but when fully developed they are special 

 folds of membrane attached to the inner surface of one or both 

 lips of the nostril and directed outwards {Trigla, Pagellus, Anableps, 

 Orestias, Clarias). 



Finally, there are cases in which there is one nostril only 

 present [Gastrosteus (Solger) ; Chromidje, Labridje (Milne - 

 Edwards)]. Or the nose may assume (apparently as a secondary 

 modification) the condition of a widely open pit without defined 

 nostrils at all (Scombresocidae) ; or by the elevation of its floor 

 and the subsequent ruptui-e of the bridge between the nostrils, 

 the cavity may be everted and the organ transformed to a bifoliate 

 tentacle (Tetrodons, Wiedersheim and Tate Regan). 



Turning now to the rosette, it is noticeable that the axis is 

 most frequently linear and continued to the front lip of the 

 anterior nostril, as in the Haddock, the rosette being usvially of 

 an oval form (Bateson's type 3) (Rosette, Column I.). 



In Ci/clojjterus, -Bovichthys, Gottus, Esox, Orestias, it is a central 

 circular boss from which the laminae radiate in all directions, the 

 rosette being circular (Bateson's type 3) (Rosette, Column III.). 



In the Eels, and to a less extent in the Siluroids and the Sole, 

 the axis is lengthened and the laminae are set in parallel series at 

 right angles to it (Bateson's type 2) (Rosette, Column II.); and 

 finally in Ophiocephalus, Rijypoglossus, and Pleuronectes, the axis 

 lies transverse to the internarial line and the laminfe are attached 

 to its posterior border in parallel series (Bateson's type 4). 



In the degenerate nose of Lopluus the few laminae left are also 

 arranged parallel to one another and to the narial axis, and in 

 the Percesoces laminfe are entirely wanting. 



Considerable differences are apparent in the shape of the 

 individual laminae of the rosette. Starting from the type pre- 

 sented by Gadus as a centre (Rosette, Column Y.), one line of 

 variation leads by the suppression of the peripheral pai-t of the 

 lamina and the exaggeration of the linguiform process (Rosette, 

 Column YI.) to the claw-like shape which is particularly charac- 

 teristic of the rosette of the Salmonidae and Clupeidae. A similar 

 though less pronounced modification is shown by the sharply 

 convex laminae of Mugil, Perca, Pagellus, or Sj)hyrcena, and pro- 

 bably the triangular lamina3 of the Eels should be included in 

 the same sex'ies. On the other hand, suppression of the linguiform 

 process (Rosette, Column YII.) gives rise to a gently curved or 

 straight free border such as that seen in Mormyrus, Clarias, Esox, 

 Orestias, some Pleuronectids, Ophiocephalus, Bovichthys, Trigla, 

 and Gyclopterus. Except in the case of the Salmons and Herrings 

 the form of the laminae has apparently little dependence on 

 natui^al affinity, except in quite closely related foi-ms. 



Yariations in the position of the olfactoiy chamber relative to 

 the framework of the face are very exceptional, being found only 

 in Motella, Zeus, and Lophius. Its shape also is very constant, 



45* 



