In 1913, a man named C.E. Inglis looked at a thin plate of glass (?) with an elliptical hole in the middle in a new and different way. Theoretically, the plate was infinitely large and the hole very small in comparison. The plate was pulled at both ends perpendicular to the ellipse.

He found that point A, at the end of the ellipse, was feeling the most pressure. He also found that as the ratio of a/b gets bigger (the ellipse gets longer and thinner) that the stress at A becomes greater and greater.

He also found that pulling on the plate in a direction parallel to the ellipse does not produce a great stress at A. This leads to the fact that a load perpendicular, not parallel, to the crack will make it grow.

Then he looked at other plates with not-quite-elliptical holes, like these. From looking at these he realized that it's not really the shape of the hole that matters in cracking. What matters is the length of the crack that is perpendicular to the load and what the radius of curvature at the ends of the hole is. The longer the hole (or crack), the higher the stress, and the smaller the radius of curvature, the higher the stress.

Then, in the 1920's, a man named A.A. Griffith extended Inglis' work. Griffith thought again about the infinite plate under tension, but he stretched the ellipse out into a crack. He did some experiments too.
His first experiment was on a soft iron wire, 100 inches long and 0.028 inches in diameter. He pulled on this wire until it yeilded. Then he took another wire with the same dimensions and put very small fractures, scratches, on it spirally. He pulled on this wire until it yielded. The scratched wire had only a quarter of the strength of the unscratched wire. He concluded that the scratches increased the stress in the wire by 3 or 4 times which made it yield earlier.

He also tested a thin plate of glass, pulled in tension, with a crack in the middle perpendicular to the load. He found that the stresses at the ends of the crack were very high and the crack weaked the glass significantly.

From these two tests he concluded that materials that are fractured, no matter how small the fracture, act much differently than the same material that is without cracks!

Griffith also introduced the notion of energy sources and sinks in crack propagation which was mentioned earlier. He said that for a crack to grow, it was necessary for their to be enough potential energy in the system to create the new surface area of the crack. He did not know that it takes more than this for a crack to grow though