Yes - each purple step is a ‘shortcut’ directly to the sequentially numbered steps (with trigs or not) that are present on the ‘normal track’
So if your EUC seeding has 4 trigs playing, it’s not rotating through four trigs from that track, then the next four and so on - it’s literally that the first euclidean active step will play the first step position on the 'normal pattern (trig on it or not, if not then track sound used)
So you always start from the left and work through the steps, the only way to reach and loop from the end is for every euclidean seed trig to be active - so it’s just a linear playthrough - you are not scanning and looping through teh underlying steps, you are playing specific steps from the underlying pattern, from the beginning (unless rotated of course)
This will make sense when you ‘print’ a euclidean pattern back into position or use p-locks on euclidean steps – any apparent ‘gaps’ between euclidean steps are non-existent on the underlying pattern … if you p-lock the first EUC step, you are actually p-locking step 1 from the pattern, the next EUC step is step 2 from pattern (again, if not rotated)
There’s therefore a consistency with what you are acting on in real terms … you are just windowing in on a part of it and changing the temporal position of the elements in the window - it means you can shrink and grow the window but it’s always stored temporarily in the pattern as a different ‘normal’ pattern (until printed)
I guess you can linearly step through the underlying p-locked pattern if you had a one pulse and rotated after each step is played, this way the window is moving step by step along, but this is not an automate-able thing - in any case, the possibilities with all the other parameter locks and TRCs etc is mindboggling - it all makes more sense after a few plays