620 PROCEEDINGS OF THE ACADEMY OF [Sept.. 



assuring them in a touch of the antenna? that they were nearing a 

 foothold. 



Again, all the swimmers, with or without a burden, crossed the 

 channel at right angles to its length. The swimming ants varied in 

 length from five to seven millimeters, but the longest ones rarely re- 

 passed the channel, and the transits were made chiefly by ants from 

 five to six millimeters long. They therefore had to swim at least twice 

 their length after relincjuishing one shore before they touched the other. 

 They always lingered long on the brink, running to and fro; entered 

 the water with much hesitation, and relinquished the shore with delay. 

 I supplied as much water as was lost by evaporation, and kept the 

 surface of the channel nearly level with that of the island and the 

 edgeway. The concavity of the meniscus was slight, but capillarity 

 always hindered the ant in her departure at one shore and somewhat 

 accelerated her arrival at the other shore. 



My first experiments were made with a view to ascertaining whether 

 the ants laid each an individual track across the water, as is always 

 done in traversing a solid. 



Test a. — I selected ants that must have crossed the water at least 

 once, because they were returning to the island, and when they had 

 released their hinder feet from the edgeway, I took, with a pipette 

 having a curved point, some drops of water from the surface of the 

 water directly in front of the swimmer. (_)f thirty-one ants, twenty- 

 one went immediately on their way to the island, and ten turned 

 back to the edgeway. The taking of the water may not have caused 

 all of the ten to turn back, since many other ants, finding the 

 struggle of crossing too great, turned back when the water was 

 untouched. Had the ants that continued on their way had a track 

 previously laid on the water, it would probably have been destroyed 

 by my action, and the twent3^-one ants would also have turned 

 back. 



Test h. — Choosing an ant that was returning to the island, while 

 she was in mid-channel, I passed a knife-blade five millimeters broad 

 several times around the ant, penetrating the water to half its depth. 

 Often ants thus encircled none turned back. 



Test c. — I swept the floor of the room violently, raised much dust, 

 and left the channel imtouched during the ensuing twenty-four hours. 

 I then counted the ants that crossed the channel during a half-hour, 

 and found that twenty-one passages were made to the island and 

 twenty-five away from the island. The latter number included four 

 ants with burdens. With a shred of cloth I then brushed the dust 



