This section is from "Scientific American Vol II. No. 1", by Munn & Co. Also available from Amazon: Scientific American Science Desk Reference.
Our correspondent S. B. cannot comprehend that the strength of iron for a cylindrical boiler should be in direct proportion to the diameter thereof, in order to sustain an equal pressure per square inch; wherefore, we must reason with him on the long scale. The cohesive strength of good iron is 64,000 lbs. per square inch; and of course, a strip of boiler-iron plate 1/8th inch thick will sustain 8000 lbs. If a boiler made of thin iron is 14 inches in diameter, or 44 inches in circumference, each inch of its length will contain 44 square inches, and either half thereof will contain 22 inches, and as the pressure on this portion is sustained by at least two inches of width of plate,--one inch on each side,--it follows that it will sustain a pressure of at least 700 lbs. per square inch, in the direction of circumference. If the diameter is double, the number of square inches will be double, and will require double the thickness to sustain equal pressure. With regard to the pressure endwise, the area of a cylinder head 14 inches in diameter is 154 inches, and the strength of the 44 inches of circumference would be sufficient to sustain 352,000 lbs., which, divided by the area, is 2,275 lbs. per square inch.
If the diameter is 56 inches, the circumference being 172, would sustain a pressure endwise of 555 lbs. per inch. Thus it will be seen that if the cylinder were even 20 feet in diameter, the iron would better sustain the pressure on the head that on the periphery. With regard to the requisite strength of the cylinder's head, if they are made in a semi-spherical convex form, they will require no more thickness of plate than the cylinder: but if they consist of plane disks, the thickness thereof should bear the same proportion to that of the periphery that the area in square inches does to three times the circumference. But in general, no other rule is observed for the thickness of the heads, than to make them extravagantly heavy, without much regard to theoretic calculation.
 
Continue to: