260 THE NATURAL HISTORY 



way ; for they cause a glass bowl to be blown with a 

 large hollow space within, that does not comnaunicate 

 with it. In this cavity they put a bird occasionally ; so 

 that you may see a goldfinch or a linnet hopping as it 

 were in the midst of the water, and the fishes swimming 

 in a circle round it. The simple exhibition of the fishes 

 is agreeable and pleasant ; but in so complicated a way 

 becomes whimsical and unnatural, and liable to the 

 objection due to him, 



" Qui variare cupit rem prodigialit^r unam." 



I am, etc. 



LETTER LV 



October 10, 1781. 

 Dear Sir, 



I think I have observed before that much the most 

 considerable part of the house-martins withdraw from 

 hence about the first week in October ; but that some, 

 the latter broods I am now convinced, linger on till 

 towards the middle of that month : and that at times, 

 once perhaps in two or three years, a flight, for one day 

 only, has shown itself in the first week of November. 



Having taken notice, in October 1780, that the last 

 flight was numerous, amounting perhaps to one hundred 

 and fifty ; and that the season was soft and still ; I was 

 resolved to pay uncommon attention to these late birds ; 

 to find, if possible, where they roosted, and to determine 

 the precise time of their retreat. The mode of life of 

 these latter hirundines is very favourable to such a design ; 

 for they spend the whole day in the sheltered district, 

 between me and the Hanger, sailing about in a placid. 



