The Imperfectionists

Chapter 3Occurrent Behavior

Let's Weigh the Money

A story about experimenting on a high-stakes problem for the Federal Reserve Bank has stuck with us for years. Two of our former colleagues, Ted Hall and Don Watters, explain:

“We found that there were hundreds of people who worked six-hour shifts (at the Fed) counting money. They would open the packs received from commercial banks and recount them. The frequency of over- or under-counts in those packs was very low.”¹

Ted and Don considered the costs of this time-consuming approach and asked themselves what the alternatives were. Counting machines were at least a decade away. Then Don had an epiphany: On a visit to San Francisco's Wells Fargo Museum, he recalled seeing the precision scales used for weighing gold during the California Gold Rush, and wondered if the bills in the Federal Reserve could also be weighed with such a device.

Ted and Don decided to conduct an experiment. They set up two bundles of paper currency in the vault, arranging for one to be counted by humans and the other packets to be weighed against the standard weight they had determined. “Then, using statistical sampling, we analyzed which system was more accurate at identifying the overs and unders.”

Arthur Burns, chair of the Federal Reserve, was on hand to observe the test and see the results. Initially, both the “piece-verified” pallet (the one that had been counted twice by hand) and the second pallet, which was weighed and statistically sampled, ...