ITHEORY OF WALLS. 99 



SCHOLIUM. 



In eftlmating the value of w in thefe computations, 

 1 have fuppofed it to be \ the weight (W) which 

 would fuftain the wall by vi^hich the experiment is 

 made, /;/ equilihrio : this I fufpeft is below its value ; 

 particularly if the mafonry be old. In afcertaining the 

 value of w I would undoubtedly make various experi- 

 ments with mafies of mafonry from one to tour or five 

 years ftanding, fo as to compute for works whofe 

 walls may probably remain for fo many years before 

 they be clofed up with earth. If the mortar be very 

 good, the cohefion of a wall well built and feafoned, 

 muft become a very important obje(5l in the conftruc- 

 tion of large fortified places ; — for by that the expence, 

 as well as time and labour in the building, muft be 

 confiderably reduced. To ufe no more materials than 

 what are neceffary lliould be a maxim in fortification, 

 but then to determine the exa6t dimenfions of any 

 particular work, fo that it may anfwer the purpofe 

 intended, and yet have no ufelefs materials about it, 

 muft require a mathematical inveftigation before any 

 rule can be obtained for proceeding upon folid and 

 infallible principles. Engineers, to whom the direc- 

 tion of the mod important works of a nation is in- 

 trufted, ought to be capable of determining what is 

 precifely neccfTary to be done in all cafes, that no ufelefs 

 expence may be incurred, but inftead of having re- 

 courfe to fcience, men in general, depend upon what 

 they call experience ; forgetting that in praiSlice alone, 

 there are no means for drawing general conclufions ; 

 but that wc obtain, from ex/icriment, the requifite data, 

 to reafon and gcncrcilife upon, and by fuch materials 

 we are enabled to build a theory, to which practice 

 muft be ever fubordinate and conformable. Should 

 any circumftance occur in the courfe of pra6lice that 

 has not been confidered in that theory, fuch circum- 

 ftance fliould be then taken into confideration j but 

 let no conclufions be drawn from thence, but what 

 are correft and fcientific. For to attempt to reafon 

 without principles j to fubftitute hypothefis for fa8:s, 



G a and 



