Please leave the silly 6 weeks claim [Robert Graves] to fiction.
I quote the late Martin Gillott, my good friend, army officer, stockbroker, engineer, mathematician, historian:
The Six Weeks claim, most famously recycled and promoted in John Lewis Stempel's book of that title claims the stats related to the 'bloodiest periods' of the war.
I believe nothing was more bloody than 1914 from the perspective of a Battalion Officers. The Infantry's 1914 Star medal roll captures 5,462 Regular Battalion Officers (not including those on the Staff. Of these some 1,829 became fatal casualties - almost exactly one in every three - and are listed in ODGW. Using the disembarkation dates and their fatal casualty dates it is fairly easy to reconstruct the 'average' time these Officers served before becoming a fatal Casualty
23 weeks and 4 days.
If we limit the study to the 3,201 Subalterns (Lts and 2 Lts represent 59% of all Battalion Officers on the rolls) some 1,145 ultimately became fatal casualties (36% fatality ratio in case you are wondering). The Subalterns of this cohort who died, served from disembarkation to death for an average of...
28 weeks and 4 days
Large sampling indicates the Officers of the 1914-15 Star cohort fared slightly better, although there are large skews depending on which theatre they started in. The fatality ratios and length served before becoming a fatal casualty for those who started on the Western Front is broadly longer to the 1914 Star cohort. I think it simply tells us that being a subaltern in 1915 was just as hazardous as 1914 (similar proportions died) but the later cohorts who died, served for longer before meeting their maker. It does also prove the Six Weeks argument, even if restricted to just those who died, is nonsense. At best it is out by a factor of four.