The 2003 FMT

In 2003, conditions weren't as good as the previous year (there was a major solar flare that disrupted propagation quite badly), and I copied only the two lower bands:

Band N8UR Measurement ARRL Measurement Error
80M 3 585 383.685 3 585 383.7 -0.015Hz -- -4.2x10-9
40M 7 050 409.949 7 050 409.9 +0.05Hz -- +6.9x10-9

If you're interested, here's my ARRL Submission Letter (pdf).

The receiver used in 2003 was an HP 3586C Selective Voltmeter, which is often available on eBay for a couple of hundred dollars, and is a really useful piece of gear for HF measurements of any sort. The reference signal came from the 3586C's built-in tracking generator, which is very convenient because it creates a single-adjustement system -- tuning the receiver automatically tunes the reference so that it always remains precisely at the center of the receiver passband (as you tune, the reference frequency note stays constant at about 1850Hz, while the unknown signal's note changes with tuning). Tuning is a matter of using first the ear, and then the spectrum analyzer, to place the unknown signal's note just slightly below that of the reference.

The 3586C's external reference input, which controls both receiver tuning and the tracking generator, was hooked to an HP 5065A Rubidium frequency standard that was being monitored against LORAN-C -- the frequency offset was about +1x10-12.