Practical information on constructing and repairing knife edge suspensions:
Laurie Penman's Clock Design and Construction and Practical Clock Escapements have useful information on knife edge suspensions.
Materials for knife-edges and pivots:
Traditional materials used for pendulum knife edges are steel and agate. Some modern clocks with knife edge suspensions make use of tungsten carbide, sometimes formed to a radius using grinding or EDM techniques.
The Granta Materials Intelligence web site discusses material choices for knife edges from an engineering point of view. The Granta site suggests that elastic and plastic deformation are important varibles to control in a knife edge suspension. Elastic deformation is minimised by choosing materials with high Young’s modulus; plastic deformation is limited by choosing materials with high hardness. Granta suggests making a graph with Young’ modulus on one axis and and hardness on the other and pick the materials with high values of both. "The very best are all ceramics: boron carbide, silicon carbide and tungsten carbide. If the selection box is relaxed so that the first metals appear, the selection picks up medium carbon steel, high carbon steel and low alloy steel. All are sensible choices: the ceramics when the ultimate precision is required, the steels when robust design able to deal with shock loading is needed."
Knife edge theory and experiment:
Philip Woodward discusses large radius suspensions in Roller Suspension, a Tangled Tale, 1,& 2 in the Horological Journal and reprinted in the book Woodward on Time.
Blair Martin's thesis (1962) An Analytic Investigation of the Ultimate Limitations on the Accuracy of the Pendulum Clock contains a detailed analysis of knife edge suspensions on pages 33-39.
On the effect of the rounded edge of a prism on the oscillations of a pendulum by A. A. Korobitsin
Knife-Edge Bearings by P. J. Geary is a bibliographical survey of knife-edge bearings. This is a document published by the British Scientific Instrument Research Association in 1955. It is probably the most complete review of knife edge theory and experiment up to that time but it is not focused on horological applications.