It was very common in text books when showing nested loops

int nWhatTheCount = 0;
for (int i = 0; i < 10; i++) { 
    for (int j = 0; j < i; j++) { 
        for (int k = 0; k < j; k++) { 
            for (int l = 0; l < k; l++) { // and on, and on
                nWhatTheCount++;
            }
        }
    }
}

j is for a loop in a loop.

foreach is useful when you don’t need to know the index of something. If you do, conventional i, j, k, etc. are useful.

A lot of it depends what you’re doing (number crunching, for instance) or if you’re in a limited programming language (why won’t BASIC die already?) where parallel arrays are still a thing.

two letters it is then

And people who iterate over 3D space using firstDimensionIndex, secondDimensionIndex, and thirdDimensionIndex instead of x, y, z have no sense 😜

If you have the need to nest 5 levels of for-loops, I suggest taking a step back and rethinking your approach, my friend.

Even if that other approach is just refactoring it into separate methods.

I just do i2, i3, etc

When you have multiple indices you’re also bound to have multiple cardinals those indices count up to, say foo.length and bar.length, so foo_i and bar_i are perfectly legible and self-documenting. A bit Hungarian but Hungarian is good in small amounts. Unless you’re dealing with width and height in which case it’s x and y but it’s not that width_i would be incomprehensible.

Two or three "i"s is readable, but any more and you’re counting.
I’Ve started using i, k, m, n that’s usually enough.

At this point we might as well go full Roman as you suggested. MXMCIIV to MXMCCVII as indices.

due to convention everybody understands what i and j are, I don’t think they need longer names. If it’s something more complicated than a counter or index then maybe you should be using a foreach loop instead (if language supports it)