1903.] NATURAL SCIENCES OF PHILADELPHIA. 621 



from the whole surface of the channel, being careful not to touch its 

 shores. I then counted all the ants that crossed during the ensuing 

 half-hour. Twenty passages were made to the island, and twenty- 

 two were made from the island. The latter number included two l^ur- 

 den-bearers. The close correspondence of nimibers is explained by 

 the fact that a few individuals among the ants do most of the work 

 undertaken. 



After the removal of the dust, with displacement of the particles 

 of water over the whole surface of the channel, there was among the 

 ants no action indicating either a difficulty in pursuing their routes, 

 or a necessity for laying new tracks across the channel. 



Test d. — I next considered only ants that had taken up a burden 

 to carry away from the island, and when I saw one of these about to 

 enter the water, I swept its surface with a little cloth broom that 

 extended nearly across the channel. Fifteen burden-bearing ants 

 crossed over water whose particles had just been disturbed to a depth 

 of several millimeters. Their course was as direct and their time 

 of crossing was as brief as that of l^urden-bearers whose path had 

 not been thus swept. 



These tests show that the ants did not depend on a previousl}^ laid 

 track when they crossed the water. They evidently expected to find 

 the water in the place where they had previously encountered it. 

 Each ant pursued her laid track on the wood to the edge of the channel, 

 and then orienting herself by something other than her track, she 

 crossed and picked up her track on the opposite shore. 



Bethe's well-known hypothesis, that the ants orient themselves 

 through a polarization of the scent laid down by their feet, is surely 

 inapplicable when the ants have no track, and keep to their bearings 

 without one. 



■ The difficulty of crossing the channel was greatly increased when the 

 ant was forced to turn herself in the water in order to pursue her 

 route, but even under this arduous strain most of the ants maintained 

 their course. 



Test e. — With a small knife-blade I swirled the water when the ant 

 was in mid-channel, so as to turn the swimmer once or more around, 

 leaving her with her head toward the shore from which she had just 

 come. The results were as follows: 



1. Of nineteen ants, without burden, on their way from the edge- 

 way to the island, fifteen turned themselves in the water and swam 

 to the island. Four failed in the effort to turn and went back to the 

 edge way. 



