4 8 
SEA SAND. 
arrangement than appears at first sight. Let us make a 
diagram showing by the heights of vertical lines the quantities 
of sand in each jar in order, 
and then imagine a curve drawn through the tops of the lines. 
Now a curve of this volcano-shaped form happens to be very 
celebrated. It is well known to mathematicians. A rudi¬ 
mentary knowledge of it follows from the binomial theorem, 
but its properties are more completely discussed in the integral 
calculus. It turns up in all sorts of investigations. This 
experimental illustration of a general law has been worked out 
as one of the most elegant instances which I can find to 
illuminate a principle of far reaching significance. 
Consider what possible resemblance there could be between 
the following sets of statistics—(i) the marks of candidates for 
admission to the Royal Military Academy at Woolwich, (2) 
the statures of men in inches, (3) the erroneous results of 
schoolboys’ experiments on latent heat, (4) the number of peas 
in a pod, (5) the diverse lengths of guillemots’ eggs. But in 
everv one of these cases the statistics can be thrown into the 
•/ 
form of this curve. For instance, the vertical lines (or more 
exactly the areas between them) may stand for the numbers of 
examination candidates whose marks fall within certain limits. 
The curve comes to us by the application of the theory of 
probability to the errors of astronomical observations. It has 
been called the curve of distribution of errors. Mr. Francis 
