118 



line h in some planes through b and single in other planes through h. 

 To prove this it obviously is sufficient to show that lohen a point 

 of intersection A counts double on a line b in a sequence of planes 

 tfj, r5>3 . . . through b, converging toivards a limiting plane (^, then A 

 also counts double on b in r)\ 



Let us imagine two parallel planes, also parallel to b, situated 

 close to b and on different sides of that line. The lines of intersection 

 with r^j, f^j . . . . (^ are respectively denoted by i^/ b,' . . . . b' and 



b^" bj" b". Now if the above proposition were false, then in at 



least one of the two planes, for instance the first, there would be 

 every time tfoo points of intersection with //„, converging together 

 towards one point of intersection with b'. This would remain the 

 same when the plane, parallel to itself, moves towards b. But then 

 it is unavoidable that two branches departing from A in S„ which 

 keep finite breadth, converge towards one single finite branch 

 departing from A in (), hence the two sectors of the surface, meeting 

 at that branch, would be situated on the same side of () and this 

 has been shown to be impossible at the begiiming of § 3. 



