( xxxiv ) 



astonishingly active and expert at getting out of the way of 

 ants that come up behind them. They rarely settled outside 

 the track of the ants. I concluded that one or two that I 

 saw in that position were, for the time, replete. When I 

 brought a little glass tube up to them — to within a quarter 

 of an inch, and not with any great caution either — they flew 

 an inch or so over the stem of the tree, and if their flight took 

 them once more amongst the ants, they simply, as it were, 

 watched points,' taking care not to let an ant get them 

 from behind, but making no attempt to importune them as 

 they passed. I saw several, presumably hungry, flies alight 

 right in the ' busiest ' part of the track and immediately 

 begin w chivying ' the ants — that is really as good a descrip- 

 tive term as I can find. An ant coming up to meet a fly 

 would dodge to the side. The fly immediately turned round 

 and ran after it. Their actions were extraordinarily like 

 those of the importunate beggar. If the ant took no notice, 

 further than to keep on dodging out of the way, the fly didn't 

 waste much time, bi;t turned round and importuned another. 

 Those that refused to be ' bled ' or ' touched ' seemed to show 

 no serious resentment and made no attempt to drive off the 

 beggar by force. Cremastogasters running up and down a 

 tree are constantly making little regurgitory exchanges, a 

 momentary ' osculation,' and each hurries on its way. If 

 anything the fly and ant exchange lasted rather longer, but 

 the ant continued on its usual hurried way just as if it had 

 met one of its own kind. It is just the same when Harpa- 

 </<>nu/ia is the other part}' in the transaction. Harpag&myia, 

 however, hovers an inch or less over the line of ants (at times 

 resting on the stem and dodging out of the way when neces- 

 sary), till it sees what is presumably a likely ant. If the 

 ant is running downwards the mosquito drops down (in 

 flight) also, keeping a little in front of the ant — as near as 

 possible without touching it. The ant tries to evade it, but 

 the mosquito as a rule declines to be put off, and the ant at 

 last stops. The mosquito quickly settles and the usual rapid 

 exchange begins, the mosquito thrusting out its proboscis — 

 which when not in action is carried bent under the body 

 much like the rostrum of a Reduviid bug so that the swollen 



