366 SMITHSONIAN MISCELLANEOUS COLLECTIONS [VOL. 50 



recent works on this subject by Roche (1891, p. 4), Baer (1896, p. 

 454), and Bertelli (1900, p. 45). A list of the literature relating to 

 the subject is given at the conclusion of this paper. 



Methods ' 



The investigations were made upon our ordinary semi-domesti- 

 cated pigeon (Columba livia). It is by no means easy to demon- 

 strate the air-sacs with all their diverticula in their true shape and 

 size. Most authors who have studied the morphology of these 

 organs have employed the ordinary injection-method, that is to say, 

 they injected fresh specimens. Although this method, when judi- 

 ciously applied, may give useful results, it is not free from defects. 

 If the air-sacs are filled with a suitable stiff-setting mass, it is possi- 

 ble, by dissecting out the injected material, to obtain a cast showing 

 the shape of the cavities. Unfortunately it is impossible to expel all 

 the air from the sacs before injecting them, so the injected mass 

 can fill only a portion of the cavities, the spaces occupied by the re- 

 sidual air appearing as hollows in the cast. It is true that the mass 

 penetrates well into the large sacs. These, however, are easily 

 enough demonstrated without injection. Into the smaller diverticula 

 and secondary passages which are hard to find by mere dissection, it 

 penetrates only partially, or not at all. Some authors, as Roche 

 (1891, p. 17) and Plateau (1890, p. 72) have got rid of most of the 

 residual air by means of complicated systems of tubes inserted into 

 the air-sacs ; but this method also has its disadvantages. In life the 

 air-sacs are enlarged by pressure on their walls varying locally with 

 the nature of the surrounding structures and the size of the sac; 

 when injected, the walls are subjected to a uniform pressure. The 

 expansion of the various parts of the air-sac system produced by 

 injection is therefore very different from their normal expansion 

 in life. 



Results of injections vary also according to the position of the 

 bird during the process; if it is injected when lying on its back the 

 air-sacs are filled in a different manner than if it is injected when 

 lying on its ventral side. It may well be that this unreliability of 

 the injection method is responsible for the mistake made by Baer 

 (1896, vol. xxi, fig. 3a) in representing the lung of the pigeon rela- 

 tively much too small, a mistake which appears to have markedly 

 influenced his views regarding the mechanical work of breathing in 

 birds. 



An injection-mass cannot be forced in without using considerable 

 pressure. This causes an abnormal enlargement of the thoracic and 



