While listening to some WSPR signals (and testing my new encapsulated WSPR beacon) I heard some signals that were throbbing at about the baud rate. Instead of a more or less constant carrier, I’d here them swell and fall, swell and fall repeatedly. When I generated a spectrogram, I got this:

You can see a copy of the original signal about 4 Hz higher than the original. Is this an example of multipath?

Addendum: ZL1BLU has a nice page on frequency measurements which describes a lot of the kinds of distortions you might expect to see over an HF path. If I read this correctly, the problem is actually caused by Doppler shifts as the index of refraction in the ionosphere varies over time, causing the frequency to shift over time. If you look carefully at the spectrogram taken of the 16Mhz VNG transmission, you’ll see that the received signal varies quite a bit up and down, but also splits rouhgly 1/3 of the way through the signal into two parts, and again at the end. You can also see the keying sidebands, which are symmetric (both above and below). Interesting stuff.