Burke's Peerage has a very confusing lineage at this point. I will give all four "generations". The 2nd generation is apparently older than the first--so I don't know if Burke's is actually indicating four generations. They are certainly indicating confusion.
-----Text Copied from Burke's Peerage indicating 4 or 6 generations-------
-----of which generations 2-5 are in question (#'s added for clarity)---------
1. Sewallis de Shirley's 2nd son:
2. Henry de Shirley, of Shirley; living (of age by?) 1195; married 1205 Joanna, daughter of John de Clinton, of Essex, and had, with a younger son (Ralph) and a daughter (Avice, married her cousin Serlo de Monjoye and had issue):
3. Sewallis de Shirley, of Shirley; living 1167; married Isabel (married 2nd Joscelin de Nevill and 3rd Ralph Musard of Staveley), daughter of Robert de Meynell of Meynell Langley, and died in or after 1228, leaving, with a daughter (Elizabeth):
[end of page 1050, and continuing on to top of page 1051]
4. Henry de Shirley, of Shirley, had:
5. Sewallis de Shirley; Coroner of Derby 1242, as which is recorded as being "insufficiens" (i.e. Not up to the job? or perhaps overworked with too many cases); had:
6. Sir James Shirley; married 1st well before 1280 Agnes--and had a daughter (Isabella married c1296 William de Yeaveley); married 2nd by 1280 Isabella, widow of John de Brunston, and died in or after 1296, leaving by her.
------------------------------------------------------------------------
I believe that it is pretty obvious that generation #2 is a good fit as parent of #5 Sewallis Coroner of Derby in 1242. What I don't know is if Burke's Peerage is stating that. One could interpret that Burke's Peerage in #4 is making a backward reference to #2. I think that it is also pretty obvious that #2 is not a parent of #3, since someone "living 1167" is not very often a child of a 1205 marriage by a father who might have been of age by 1195.
In order to make the best fit for the above, I am switching generations #2 and #4 in the above, trading a Henry who obviously belongs in the #4 slot for a Henry about whom nothing is known. Thus making it #1, #4, #3, #2, #5, #6, which at leat preserves the order of "Sewallis" and "Henry" names in the lineage. An alternative would be #1, #3, #2/#4 combined, #5, #6. Another alternative is #1, #2/#4 combined, #5, and #6 leaving out #3 which does not seem to fit.