How much can we trust our understanding of how things work? There was a special joint web ex meeting this week between two AGMA committees that got me thinking about how we come to accept some methods and, over time, lose sight of how they came to be. The meeting came about because the Wind Turbine committee is working on a new and improved version of ANSI/AGMA/AWEA 6006 and an interpretation was requested from the Helical Gear Rating committee on the use of reliability factors. The Wind Turbine community has been putting a lot of resources into improving turbine reliability and is ready to greatly modernize and expand the calculation of system reliability estimates. The reliability factors in AGMA 2001 modify capacity based upon the designer’s tolerance for failure. If 1-failure-in-10 can be tolerated, for example, the gearset can transmit more power than if only 1-failure-in-a-1,000 is acceptable; the default setting is 1-failure-in-100. In response to the inquiry, the helical gear rating committee checked their records and traced the chart to a 1979 meeting. According to the minutes of that meeting, the data was the result of metallurgical testing by the United States Navy. Further digging turned up some computer printouts of the tests — but no narrative of the experiment’s design or execution. This printout will now be used by the Wind Turbine’s experts to review the type of distribution that will be employed to improve the reliability calculation. An effort will be made to locate more information on the test itself so that future committee members will have a better basis upon which to build. This situation calls to mind an earlier flap over the widely used rim thickness factor. In working on the new helical gear rating standard we discovered that various product standards had different rim thickness factors, but all referred to the same Fall Technical Meeting paper from many years ago. When that paper was pulled out and studied, it was found that no one was properly applying it as it concerned a planet gear with a bearing mounted in it — not a webbed gear, as everyone erroneously recalled. The lesson I took away was that we need more traceability in the standards. No one wants to wade through hundreds of pages of footnotes in a standard, but there ought to be an annotated copy somewhere that links each piece of data or formula to its source material.