Abstract
An incrementalist view of the R&Dprocess is developed, according to which R&D consists of informed trial and error. One way of avoiding expensive mistakes is to avoid choices within the R&D program that are highly sensitive to a particular scientific claim, because a great deal of time and money may have been spent to no avail should the claim turn out to befalse. The incremental view of R&D therefore entails that no choice within any R&D program is sensitive to any scientific claim made at the time. Two tests are presented, antiseptic surgery and the first atomic pile, both of which confirm the incrementalist thesis.