I just noticed that frames 1510 収 and 1511 卑, while sharing the same primitive, actually have a very different stroke order, stroke number and appearance. I'm using the wwwjdic stroke order diagrams. That might stem from different origins (http://www.zhongwen.com/cgi-bin/zipux2.cgi?b5=%A8%F5). Too bad 収 is not in that database.

Do you think that's a big enough difference to warrant a new primitive or at least a note by Heisig? Maybe it's mentioned in the fifth edition. Have you noticed any other imprecisions like this?

I also don't like how strokes sometimes double up (measuring box) and sometimes don't (heavy).

There's a lot of primitives and radicals that have different origins but look the same. I'd be more confused if I had to try and tell them apart and give them different names. It's not like Heisig's primitives have accurate names anyway.

I'm not concerned about that either, it's just that when you compare them: (http://www.csse.monash.edu.au/~jwb/cgi-b...198_%BC%FD) (http://www.csse.monash.edu.au/~jwb/cgi-b...642_%C8%DC) then in one kanji cornucopia has 3 strokes and in the other 2.

There are quite a few of these; there are some examples mentioned in the "stroke order rules" sticky.
In cases where it's just a matter of appearance, it might not even be worth mentioning; you sort of pick that stuff up automatically. But when the difference is in stroke count, order, or direction, and Heisig doesn't mention it, I don't think it can be considered anything other than an error. I mean, by following his book as it's written, he's teaching you to write quite a few kanji wrong. This is for me the only really disappointing thing about the book, which I'm generally a fan of. I don't know whether the 5th edition has improved in this regard...

The cornucopias in 収 and 叫 are the same component. It's just that things tend to be written differently when they appear on the left (cf. 牛, 物). The cornucopia in 卑 is unrelated to the other two, but has the same stroke order as the one in 叫, so separating the two- and three-stroke cornucopias into two primitives has no valid argument on etymological grounds.

I believe Heisig mentions that cornucopia has a different stroke order/count when it's on the left and that it gets squashed when it's beneath something. It's entirely uniform and I can't think of any reason to separate it into two different primitives.

If you're looking for 収 on zhongwen.com, you need to look here: http://www.zhongwen.com/d/166/d172.htm

Heisig doesn't mention the different stroke count, so I think he made a small error there. The reasoning for separating it into two, perhaps closely related, primitives would be because that's a good way to remember it. A note would suffice though.

I don't really worry about the different stroke orders in 牛, 物 or 左,右,石　or other characters for now. My reasoning is that since actually being able to write the characters is almost overkill nowadays, the stroke order is certainly beyond what I need to know at my level.

There are plenty of primitives that lose strokes or slightly change shape depending on their location and still remain the same primitive: 食, 良, 長, 羊, 乙. If you want to make the two-stroke cornucopia a separate primitive, go ahead, but frankly, there aren't enough instances of it to merit a separate primitive.

It's not necessary to name a separate primitive for these cases, but he should mention the difference, especially when it's something as fundamental as stroke order, count, or direction. That's the issue. He *usually* does this, but he misses quite a few, too (of the ones you list, 羊 is another example). To me, this is no different than printing an incorrect stroke diagram.
Being able to write the characters is "overkill?" The book only promises two things, and one is that you'll be able to write the characters.
[in the "stroke positioning sticky, Katsuo mentions that Heisig explains about the differences in the "cornucopia" primitive in the 3rd edition, but there's no explanation in my 4th edition -- he says they're all 2-strokers with the same order]

-this is just a font issue - it's supposed to be 2 strokes - http://www.zdic.net/zd/zi/ZdicE5Zdic8FZdicAB.htm - it was 2 strokes for a long long time - someone made a mistake because the 3 strokes variation was easier for them - but it's 2 strokes