235 
liquid vacuoles and have nothing to do with the formation of gas 
bubbles as the authors just named had supposed. 
2) Does the abstraction of oxygen from the blood by 
the gas gland involve haemolysis in the capillaries of 
the rete and gland? 
JAEGER (4) having described and given reasons for the haemolysis 
of the blood associated with the gas gland and rete in. 1903, and 
BykowskI & NusBAum (1) having figured the phenomenon in the 
following year, I was naturally not surprised to find that this same 
phenomenon was apparent in two series of my own slides — sections 
of Syngnathus and Peristethus — made from material I collected at 
Naples in 1907, and I figured stages in this process of haemolysis in 
the paper before quoted (fig. 21, pl. II). Apart from the actual 
facts shown me by my own slides, there were moreover good a priori 
reasons why haemolysis should occur. One of these I have already 
mentioned, viz. the necessity for the gas gland cells to abstract the 
oxygen in large quantities and at a rapid rate from the blood and 
therefore the necessity of absorbing the main source of the oxygen — 
the oxyhaemoglobin — in place of the weak solution contained in the 
plasma. Another is the fact stated by HALDANE (3) that in order that 
ferricyanide may act upon the oxyhaemoglobin of fresh blood and split 
off the oxygen, it is necessary first to mix the blood with an equal 
volume of water “since ferricyanide does not act on undissolved cor- 
puscles”. In fact, assuming that this haemolysis of the blood occurs, 
it is obviously necessary to suppose that the gas gland secretes two 
substances: 1) a lysin (JAEGER’s “toxin”) poured into the blood to 
produce haemolysis (the efficient action of which is ensured by the 
rete mirabile, on my hypothesis), and 2) a “pneumatogen” which acts 
in a similar fashion to ferricyanide on the oxyhaemoglobin, which, 
according to the theory, is absorbed by the cytoplasm of the gland 
cells from the blood stream. However, despite the above-named 
authorities, facts and a priori reasons, I have come to the conclusion, 
as a result of the experiments now to be described, that haemolysis 
does not occur. 
I will first record the experiments made on the Pollack. I made 
serial sections (transverse and horizontal) of the gas glands and retia 
of Pollack which had been weighted for the following number of 
hours: 24 hours (three fish at least), 16 hours (two fish), 13 hours 
(two fish), 7 hours (two fish), 6 hours (at least twelve fish), 5 hours 
(one fish) and also of four unweighted control fishes. These glands 
were for the most part fixed in Mann’s fluid (distilled water 100 cc., 
