The author misunderstood what he is referring to as a "null" test. That was actually a 3 way test, and the "null" test device has an actual null result. (I'd have to re-read to see what the "null" device had and didn't, but it didn't send the EM signals it generated into the chamber.) 1 of the groups pursuing these devices had a theory of operation dependent on a slot, but the 3 way test found identical performance between slotted and unslotted versions, not between null and functional versions.
This is also an old article, with no re-examination of newer results.