Every so often the question of Morse speed comes up. It often gets comments from certain quarters about how, way back when, they first got on the air at some slow speed, and how there was always some slow code to listen to on the bands.
Clearly, times change, and on an idle afternoon I set about reducing some of the raw spots from the Reverse Beacon Network (RBN). I tried to reduce the data to show how many unique callsigns (excluding beacons) were calling CQ at various speeds. Operators will count only once each in any one group, whether they call CQ just once, or many times. Operators who call CQ at a variety of different speeds will be counted once in each of the speed groups within which they call. The reduced data files are available.
RBN produces a lot of raw data, so I picked a couple of days not quite at random. The first was a relatively quiet Wednesday. The second was a busy Sunday when a big CW contest was running. The following charts came out of it. The Wednesday file contained over 135,000 spots, and the Sunday file contained well over a million spots. I also charted data from a file I'd downloaded near the end of 2014 which contained over 170,000 spots. Its charts were very similar to those for the Wednesday file.
Inevitably there are some selection effects. Not everyone who calls CQ is heard by a skimmer. In some places skimmers cover only small parts of the bands. Other places have no skimmer coverage at all. Band conditions will also have played a part. Europe and North America, however, are fairly well covered so the samples should be reasonably representative.
RBN does rely very heavily on CWSkimmer for its input, so anything CWSkimmer can't read won't get into RBN. Any selection bias in CWSkimmer will be reflected in the data. CWSkimmer was originally designed as a contesting aid with a focus on fast CW, so perhaps it isn't very good at spotting slower calls? It's possible that the charts are just showing where CWSkimmer's limits lie. However, the slowest spots found in the raw data showed a reported speed of two words per minute, and the fastest showed a reported speed of 62wpm, so it's probably reasonably representative over most of the normal range of speeds used by amateurs. There's a chart at the end of this page which shows reported speeds in more detail.
The charts start at the 12 o'clock position with a very thin blue slice (typically less than 1%) showing operators calling CQ at 10wpm and below, and speed increases by stages as you go round clockwise, with the thin green slice (again, typically less than 1%) just before the 12 o'clock position showing operators calling CQ at more than 40wpm.
© M0LEP (Last updated August 2015.)