VOL. XXXVIII.] PHILOSOPHICAL TRANSACTIONS. 605 



also gave the description of a new instrument, proposed to remedy the princi- 

 pal objections to the former ; and further experience has sufficiently confirmed 

 him in what he advanced. But he would be glad to have it determined by those 

 who have convenient opportunities of making experiments of this kind, what 

 would be the properest diameter and weight for a needle and card, and what 

 ought to be their proportional weights to each other when taken separately : 

 regard being had that the friction be no more than what is necessary to prevent 

 the card from being too much affected by the motion of the ship. Some ob- 

 servations incline him to think, that a sea-card should not exceed 6 inches dia- 

 meter ; and that most of those generally used, are too heavy for nice experi- 

 ments, though they may be well enough adapted for common purposes. 



In March and April, 1732, the variation at Black-River in Jamaica, was very 

 accurately observed to be from 6" to 6° 5' easterly. Off the Havanna about 4-i- 

 deg. easterly. The rest of the observations he made, are exhibited in the fol- 



The instrument used was so easily managed, that unless the sea was pretty 

 rough, an observation might be depended on to about a quarter of a degree, 

 had the card performed to the same exactness. But by comparing several ob- 

 servations made under the like circumstances, as to the weather, it seems as if 

 the virtue of the needle was not always of equal strength. Sometimes several 

 observations would agree exceedingly well ; at other times the card would stand 

 indifferently any where within a degree or more of its meridian ; and this was 

 observed in several cards. Another circumstance was surprising: the card would 

 sometimes differ about 2° from itself between the morning and evening of the 

 same day ; and this difference would continue as it were regularly for several 

 days, then vanish for a week or more, and afterwards would return and con- 

 tinue as before. 



The greatness of this difference, and the near agreement betv/een the obser- 



