Ki:si'iRA r»>m oucaxs ok |)RA(;()^-l■"L^■ i.\r\'.i-.. Ooi 



wliicli t-nahk' tlu' > )])cuin<^- <>\ the i;ill-clianil)ci- lo he closed liLiIulv 

 ai^ainst hai'irful hiiuids or solids. 



We thus >ee I'nat the lar\;e respiratory rectum in Anax iin- 

 pivatoi. \ar. iiuiiti'u'uunis, our laro^esi African (h'ajj^onflv, is a 

 sini^ularly complete and w ell-pn)tected mechanism. This a])i.)ara- 

 tus, as shown in i"Airo])ean species of Anax. was studied hy Cuvier 

 and hy a dozen others since him. The general ()])inion, exi)ressed 

 hy all uj) to even the latest. Miss .\ndrews, an American, who 

 W( rked in Kioi. was that there were perha])s ho.ooo villi. This 

 1 corrected hy counting gilk and \illi ])ei- gill, arrixing at ,^.000 

 as niaxiniiiiii number of \illi. 



In the >ame family ( .l'',shnid;e ) as Anax we ha\e four other 

 genera in this country. 1 take tip two. Anacixshiui. in the same 

 sub-famil\- ( .?ishnin?e), differs markedly from Anax. in that in- 

 stead of having gills and villi, the middle of each double row of 

 gills has a tall, thin, longitudinal fold, to which the gills, much 

 enlarged, stand as buttresses. Here the distal margin of gills 

 and fold are the ftmctioning narts. 



Mcsoi/oDipliKS. the meditmi-sized green dragonfly, wdiose 

 lar\a is the most common odonate of our Natal streams, is more 

 like Anax. T'he gills here are de])ressed into flat fatty pads, 

 around whose edge arise long villi. This form has 4.000 villi. 

 In all other features both Mcscxjoniphus and Anackcshna are like 

 Anax. 1 made a survey of eight .\merican .-Eshnids in 191 5, and 

 found much detail variation, but all forms were much like these 

 three. 



If we turn now to the famil)- Libellididce. to which our 

 common red. golden-winged, blue, and variegated-winged dragon- 

 flies belong, we shall find the rectal respiratory organ practically 

 identical iii all genera. 1 choose the genus Sympetnnii, the dark- 

 red autumn dragonfly, as ty])e. The larv<T of all the Lil)ellulids 

 are broad in the abdomen, which correlates with the internal 

 features. We note first of all that the tracheal trunks send off 

 their branches to the rectum as a sort of fan-sha])ed mass, from 

 one point. The rectum lacks an anal canal, but is right at the 

 valve. The respiratory parts show a different grou])ing ; each of 

 the six longitudinal rows has a longitudinal muscle as its axis. 

 \Xc have in each row -'4 flat gills. tluimh-sha])ed in side view, 

 with lo(;ps of trachea- in the distal part, with a " cushion " on 

 one side only, and with little fat in them. In all there are about 

 280 gills. 



Our common Trifliciiiis. Liiidcitia. Orfliclridii. etc.. show 

 only slight divergences from this — namely, in the number of gills, 

 which is sometimes 3^)0. \\'e owe to Scott. 1905, and Sadones, 

 1.S95, the first good studies of the' gill-chambers of two Lihellulids. 

 My work herein has been to extend this investigation to a dozeti 

 genera. 



Here I ma\- add that the tracheation of the rectmn i^ not 

 entirely from the two luain dorsal tracheal trunks. In all Anisop- 

 tera tlie \entral third of the rectum is tracheated from \entra] 



