LIGHTHOUSE ILLUMINATION, ETC. 633 



they may be even with less rapid speeds of rotation. Thus at Jervis 

 Bay light, which is a fairly modern one, the optic revolves on 

 rollers every minute, and gives three groups of three flashes each 

 in that time. 



The metallic reflection, or catoptric system, is practically 

 obsolete, except for secondary purposes, but has been revived at 

 the new first- order German light at Heligoland, erected in 1904. 

 Here three parabolic reflectors of silvered glass, 30 inches in 

 diameter, each with an arc electric light of 34 amperes in its focus, 

 revolve round a vertical axis and give a flash every five seconds, 

 said to be of 30 million candle-power. 



There is evidently a great variety of choice in lighthouse 

 illumination, and the subject demands the careful co-operation 

 — in its practical application — of the engineer, the student of optics, 

 and the seaman. 



6.— NOTES ON SOME EXPERIMENTS WITH MODELS OF BUILT-UP 



COLUMNS. 



By G. HIGGINS, M.C.E. (Melbourne), M.Inst. C.E., M.Am. Soc. C.E., etc. 



[Plates xlv-xlvii.] 



The object of making the experiments referred to in the following 

 notes was to ascertain the possibility, or otherwise, of determining 

 by a series of tests suitable proportions and spacing for the side- 

 lacing of columns of a certain type. The aim, in the first place, 

 was to ascertain if possible how thick and wide lacing pieces must 

 be if they are not to fail before the longitudinal members fail ; and, 

 in the second place, to find how closely the lacing points — i.e., the 

 points of connection between longitudinals and lacing bars — must 

 be spaced, if the longitudinal members are not to fail by flexure 

 between the said lacing points, i.e., by secondary flexure. 



The workmanship in the models which were tested may be 

 taken to be of the same character as ordinary girder work, if any- 

 thing, a little inferior ; but the fairly consistent results obtained 

 show that there cannot have been very much wrong with the 

 building of the columns. Only one of the nine tests need be 

 discarded. 



During the appHcation of load, each column was closely 

 watched by several persons, with the object of detecting any 

 local yielding, such as might result from imperfect workmanship ; 

 but, as usual in such cases, none was revealed ; all failed so suddenly 

 that it was impossible to say which part was the first to give way. 

 Undulations moved along some of the longitudinals shortly before 

 collapse, and the stress in certain of the lacing members changed 



