In the 19409 the method of instrumental variables was introduced for u
se in the Errors in Variables Model. According to Reiersol, the ''idea
of using instrumental variables'' was introduced independently by him
self in 1941 and by Geary in 1943. For the modem reader the ''idea'' s
ignifies the use of a particular estimator which, under certain assump
tions, is consistent for the structural parameters. Geary presents som
ething resembling this but Reiersol does not. So what was the 'idea of
using instrumental variables? The paper tries to answer this question
. It sets the work of Reiersol and Geary in the larger context of work
on the EVM in the 1930s and 1940s.