GEOMETRICAL SPIDERS 103 



other end of that filament has found a wrong attach- 

 ment, therefore the line across segment x is also out 

 of parallel. To sum up, the result of the experiment 

 is this. One turn of the viscid spiral is divided in one 

 segment. Parallelism is lost in both that and the pre- 

 ceding segment. In the one the lines diverge, in the 

 other they converge. The essential fact is that, 

 the point of measurement being removed, then the 

 parallelism is lost. 



FIG. 4. Loss of parallelism resulting from division of one turn 

 of viscid spiral in two segments. 



(a) Viscid spiral divided in segments X and Y. 



(b} Result of spider's work. Loss of parallelism in both X and Z. 



Arrow marks the direction of spider's circle. 



I perform a similar experiment, but divide the spiral 

 in two adjoining segments (Fig. 4, a and b). I will 

 not labour over the details, as these the diagram 

 should explain. Again a similar sequence follows. 

 The spider measures from the wrong lines ; the 

 attachments are made in the wrong place, and the 

 perfect parallelism is lost. 



Difficulties may prevent the success of these experi- 

 ments. For when I divide a filament, little tags of 

 the spiral are left attached to the radii. Now the 

 spider in reaching forward to find its point of measure- 



