In Robert Glass's wonderful (para)phrase, a *fault of omission* is "code not complex enough for the problem."
Often these are missing `if` statements. If you wanted to classify them by "phase of introduction" (such as specification or requirements), but it's nevertheless the code that has to be fixed.
The Glass quote is from 1981, "Persistent Software Errors 
". The actual text is:
> The major finding of the study is that a large percentage of persistent software errors are instances of the software not being sufficiently complex to match the problem being solved.
(Shades of the Principle of Requisite Variety !)
I think I first published about faults of omission in a technical report during my Masters' work (mid '80s). That got reskinned as a Motorola tech report in 1990: "A Survey of Software Fault Surveys ".
More readable is "Faults of Omission ," a 2000 editorial for Better Software Magazine.