A Remarkable Martin Roost in the City of Washington 317 



In previous accounts of Martin-roosts no mention has been made of the 

 presence of other birds, but this Washington gathering is remarkable among 

 other things for the species associated with the Martins. During the period of 

 Martin occupation, and for some time before and after as well, between 100 and 

 500 European Starlings and from 1,000 to 4,000 Purple Crackles were there 

 also, using, to a certain extent, the same trees as the Martins, but chiefly 

 other trees adjacent. Still more unusual was the presence on August 31, and 

 probably on a few subsequent days, of about 200 Bank Swallows and 50 Rough- 

 winged Swallows, which flocked with the Martins and roosted in the same trees! 



It is to be hoped that so attractive a gathering will take place often in the 

 waning summers of the years that are to come. 



Bird-Lore's Eighteenth Christmas Bird Census 



BIRD-LORE'S Annual Bird Census will be taken as usual on Christmas 

 Day, or as near that date as circumstances will permit; in no case should 

 it be earlier than December 22 or later than the 28th — in the Rocky 

 Mountains and westward, December 20 to 26. Without wishing to appear un- 

 grateful to those contributors who have assisted in making the Census so remark- 

 ably successful, lack of space compels us to ask each census taker to send only one 

 census. Furthermore, much as we should like to print all the records sent, the 

 number received has grown so large that we shall have to exclude those that 

 do not appear to give a fair representation of the winter bird-life of the locality 

 in which they were made. Lists of the comparatively few species that come to 

 feeding-stations and those seen on walks of but an hour or two are usually very 

 far from representative. A census- walk should last jour hours at the very least, 

 and an all-day one is far preferable, as one can then cover more of the different 

 types of country in his vicinity, and thus secure a list more indicative of the 

 birds present. A census covering several days would really be just that much 

 more satisfactory, but, as few of our readers are in a position to take such, we 

 think it better to rule that each report cover one day only, that all the censuses 

 may be more comparable. 



Bird clubs taking part are requested to compile the various lists obtained by 

 their members and send the result as one census, with a statement of the number 

 of separate ones it embraces. It should be signed by all the observers who have 

 contributed to it. When two or more names are signed to a report, it should 

 be stated whether the workers hunted together or separately. Only censuses 

 that cover areas that are contiguous and with a total diameter not exceeding 

 15 miles should be combined into one census. 



Each unusual record should be accompanied by a brief statement as to the 

 identification. When such a record occurs in the combined list of parties that 

 hunted separately, the names of those responsible for the record should be given. 



