FISHES FOR CONTROL OF MOSQUITOES. 9 



long. In many cases instead of pens a screen was stretched be- 

 tween two rocks projecting from the shore, or a stranded log was 

 similarly utilized. In other cases little dams were built of drift- 

 wood, stones, brush, or whatever happened to be at hand and made 

 tight with mud or turf. Sometimes holes were dug close to the 

 side of a pond but separated from it by a narrow strip of the 

 bank and allowed to fill with water. In general, the purpose was to 

 accomplish the result with as little change in the existing condi- 

 tions as possible, and one of the great merits of the work in Palisades 

 Park is that an abundance of perfectly natural checks both ways 

 were always at hand. Experiments were also made in introducing 

 mosquito larvae to fishes free in ponds or confined in pens, tanks, 

 rain barrels, and aquaria, but little value is attached to these as 

 bearing on the main problem of natural mosquito control. 



Mosquito counts and other observations and collections similar to 

 those already mentioned were made periodically both within and 

 without the experimental areas. Comparison of the stomach con- 

 tents of fishes proved especially interesting. In the original notes 

 the stomach contents of each individual are recorded separately, and 

 the different objects are entered both by numbers and as estimates of 

 bulk in terms of percentages of the whole. Owing to lack of time 

 specific determinations were seldom made except of mosquitoes and 

 forms already familiar. In most cases genera were determined, 

 but frequently the records give only families or even larger groups. 

 This seemed sufficient, as the object was not to determine complete 

 and exact dietaries but only whether the fishes were eating stages 

 of mosquitoes and how the presence of other food might influence 

 their use of moscjuitoes. It was noticeable that mosquito remains 

 were found more frequently in later than in earlier examinations. 

 Probably this is due, in part at least, to an increasing familiarity 

 with the minuter parts of the larvse and pupae, which were more 

 often detected as the eye became better trained. Any error in the 

 frequency of recording mosquitoes is of omission rather than other- 

 wise. They are easily overlooked. 



Of all the experiments made during the three seasons the most 

 conclusive were those made in Palisades Park. In the vicinity of 

 Philadelphia much annoyance and delay was caused by the depreda- 

 tions of mischievous boys and vandals, who constantly meddled with 

 or destroyed the control pens and inclosures. Many experiments 

 had to be discarded for this reason, and in some localities, as along 

 Darby Creek, it became necessary to abandon work altogether. The 

 excessive rainfall during the summer of 1919 also worked havoc with 

 some experiments by causing streams and ponds to overflow the bar- 

 riers. Notwithstanding these mishaps and limitations, a consider- 

 able body of sufficiently definite results concerning several species 

 has been accumulated. 



In reporting these results it seems best to select a few typical 

 observations and experiments for somewhat detailed description as 

 affecting particular species rather than to deal with all in a summary 

 way. Other cases would merely add cumulative evidence. In re- 

 porting the stomach contents, also, some duplicate lots as well as 

 some nonsignificant ones have been omitted from the tables, and in 

 order to save space and make the facts clearer the data have been 



