I don't have a copy of the manual j, so I can't comment on whatever the appendix said regarding "exterior walls", but I still say that common sense dictates that it does not apply to spaces with knee walls. Why not ignore the insulated ceiling and enter the measurements of the roof? Same logic applies to knee walls. All we care about is the heat load on the "conditioned space". Build a 5' x 5' dog house in an attic and imagine using the attic exterior walls dimensions to size the cooling load; same principal would apply to a space surrounded by knee walls. This is the kind of crap that results in oversized a/c's.
It's the outside surface area that is used in the loss/gain calculations. Then the volume is used to calculate infiltration values. You should invest in yourself and buy a MJ then you could answer some of your own questions. How did you come up with this thought in the first place?
And especially explain to me how you would do a manual J on a room in a finished attic where the outer walls of the attic are 25' x 40', but the conditioned area is defined by knee walls of 15' x 25'? All we care about is the heat load coming into the conditioned space INTO the conditioned space. Logic dictates that the load CAN ONLY come into the conditioned space through the envelope of the conditioned space. (not counting internal loads of course). Do a manual j calc on a small insulated & conditioned dog house in the middle of a large warehouse; by your logic, you would size the equipment needed to cool the dog house based on the exterior wall measurements of the warehouse, and that is just obviously the wrong way to do it.
When doing a room by room load, measure and round up each interior wall.
When doing a block load, measure and round up each exterior wall.
Another thing to consider: