I tried it and found out two interesting things.. First, I was able to replicate it with my Hameg HM8030-6... And jitter was proportional with delay, and wasn't multiple of 5 or something... Just more delay, more jitter...
Then I switched to DG1032Z, with low jitter oscillator... And didn't see anything anymore...
So definitely a bit of frequency modulation there....
But, with DG1032Z, if I put delay to something like 200 milliseconds and watched signal at 5 ns/DIV, there it was again...
And it got me thinking.. If you trigger now, and wait let's say 200 msec it's 2000000 periods of 10MHz.. and 5ns jitter on 200msec is 250ppb ...
And then I did autocal again.. And now it is less than 2-3 ns at 200 msec, so that would be 100-150ppb....
In specification for DS1074Z I couldn't find jitter spec.. DG1032Z is specced at 200ps at 10 MHz..