Amazingly, after wondering how on earth 2006 achieved a mean CET of 16.4*C for the whole month, computing the GFS 00z values in directly gives a CET estimate of 16.4*C to 23rd, so I guess that run gives some idea as to what it takes!
Still a week left after that though, and if conditions were bang on average following 100% verification of that run (hah!) the final CET would be 15.86*C, still falling short of 2006's staggering value.
I remain in awe of that month. Same goes for July 2006. That year was pretty sweet
I think heatwaves early on in the months with temperatures >30C are a big help, something which hasn't been seen this year. The warmth has just been consistent - something which sums up most of 2014.
The warmth in the 2nd half of 2006 was definitely remarkable and October also saw some very high temperatures, not sure if this year will go the same way though.
The most remarkable months of all IMO are February 1779, May 1833 and June 1846, those records will take some beating.