Page:Popular Science Monthly Volume 6.djvu/674

654 several radii in their claws, and draw them up and let them go suddenly; such a habit may have been the foundation of the remarkable device adopted by Hyptiotes.

One further inquiry is suggested by the fact that the net consists invariably of four radii. Whatever other variations there may be in the spider's work, as to the size and proportions of the net, and the number of interradial lines, the number of radii is constant. In more than a hundred nets, I have found the number to be four, never more nor less. Now, this seems to offer a confirmation of the common idea that spiders' webs, like bees' cells, are constructed with absolute accuracy, and are models for poor humanity.

 Nephila plumipes, made in a wire frame, and photographed upon wood reduced. In nature, the free radii, as above described, occupy about $1/6$of the area; but the web of which a figure is given was made upon a frame, the limits of which seem to have interfered with the extension of the loops above the level of the centre of radiation.

But Prof. Jeffries Wyman has shown that no such exactitude prevails with the cell of the honey-bee; for, while the average diameter of a large number of worker cells is about one-fifth of an inch, yet the difference between two cells has been found to be one-fortieth of an inch, and the aggregate diameter of ten cells may differ from that of another set of ten cells one-fifth of an inch, or the diameter of a single cell. The width of the sides varies to an appreciable extent; likewise the angles between the sides; a fourth face is often introduced into the base, and the rows of cells may be greatly out of line; in short, while it is probable that the bees work with reference to an ideal or type implanted in them, Prof. Wyman is inclined to doubt whether a type-cell is ever really made.

The reader will now be prepared to hear that, after careful examination of large numbers of nets of many different spiders, I have yet to find one in which the irregularities could not be detected by the