74 VIEHilEYER. 



foundations of tbese nests consist for tlie most part of very small, irregular 

 aecmnulations of earth. The whole structure rarely attains the size of a man's 

 head and is surrounded bj' a somewhat thin, earthj', crust-like cover. Inside 

 it consists of numerous irregular chambers having no particular design. Plants 

 soon begin to grow over the entire surface of the nest from seeds previously carried 

 in by the ants. Some of these plants attain a considerable size. Still more earth 

 is now carried in by the ants so that the plants do not lack material for nutrition 

 and can develop into enormous clusters. Since in this manner the ants often 

 cultivate flowering plants, their nests may be called 'flower gardens of ants' in 

 analogj' with Moller's 'fungus gardens.' As these plants, thus bred by the ants, 

 are, by reason of their care, able to live in the air in the same manner as epiphytes 

 on other plants, I shall call them 'ant-epiphytes.' " 



In this comparison too mitch importance should not be attributed to 

 the fact that the plant was probably a true epiphyte and received no direct 

 benefit from the accumulation of earth by the ants. The relationship 

 between the plants and the ants here existing is not sj'mbiotic in the 

 strictest sense, because it directly benefits only the ants, in that the 

 interlacing of the plant roots increases the stability of the nests. The 

 benefit which the plants receive from this relationship is a very equivocal 

 and incidental one. As a rule, one does not speak of benefit, but rather 

 of injur}', so long at least as it is not proved that the ants make selections 

 among plant species. It is therefore not difficult to imagine similar 

 conditions in the case of Oamponotiis quadiisectus Smith; on the other 

 hand, it is also easily possible that the ants take possession of the in- 

 terlacing roots simply as a suitable place for their nest. I simply wish 

 to indicate here the probability of the first-mentioned association, in 

 order to call attention to these relations, further observations upon the 

 construction of the nests of C. quadiisectus being needed in order to show 

 which of the two possibilities applies to the case in question. 



The relation of the ants to the lycsenid chrysalids is, however, clearer 

 and more interesting. The pupa sent to me is about 17 millimeters long 

 and shows the closest similarity to the dominant forms in the genus 

 Arhopala. The abdominal extremity is developed into a large cupule or 

 pulvillus-like depression, presumably similar to that of many of its 

 closely related allies, at present unlmo^Ti, and by means of which it was 

 fastened to the bark of the tree within the ants' nest. Dorsallj^, on the 

 seventh abdominal segment, exactly in the place where the mj'rmecophi- 

 lous lycsenid larvae have the opening of the secretary gland, there is 

 situated an oval, chitinous, , crater-shaped cavity 1.3 millimeter in length 

 . and 0.7 millimeter in width. Its edges consist of strong, dark-colored 

 chitin and project considerably above the surrounding surface. Inside 

 the oval zone the chitin is much thinner and readily distinguishable by 

 its yellowish color, though especially so when viewed by transmitted 

 light. When the butterfly is withdrawn from the skin of the pupa, one 

 notices 'plainly that a fine transverse incision is present between two 

 chitinous lamellae, which extend into the crater from the cephalic and 



