One couple or two individuals? T 0783/09

The famous german writer Goethe has reasoned on the leave of the Gingko tree:

 
"Is it one living thing
That has become divided within itself?
Are these two who have chosen each other,
So that we know them as one?

I think I have found the right answer
To these questions;
Do my songs not make you feel
That I am both one and twain?"

Similar thougths may have arrived to the technical Board of Appeal in T 0783/09, pointed out by Laurent Teyssèdre, which has considered a case limiting the rule that a selection out of two lists of else non-individualized features is considered novel (T 12/81). 

In the case, a paragraph 21 of the specification reads:

In a very preferred embodiment .. the  DPP-IV-inhibitor is selected from LAF237 and DPP728 and the further antidiabetic compound is selected from a grop consisting of .... [22 individually mentionned compounds follow]".

The board then argues: "Thus, the fifth paragraph of page 21 indicates two individual DPP-IV inhibitors, .... and twenty-two individual antidiabetic compounds ..." and "Thus, the skilled person would directly and unabiguously recognize forty-four individual combinations, among them the three "basic" combinations referred to in claim 1."

As to the reasons for deviating from T 12/81, the board says:
However, given the term "can" in the citation from
decision T 12/81, the absence of a direct and unambiguous disclosure for individualised subjectmatter is not a mandatory consequence of its
presentation as elements of lists. Thus, the "disclosure status" of subject-matter individualised from lists has to be determined according to the circumstances of each specific case by ultimately answering the question whether or not the skilled
person would clearly and unambiguously derive the subject-matter at issue from the document as a whole (reasons, point 5.6).

We would be curious to learn what circumstances of this specific case have led to this surprising answer. Is it sufficient that one of the lists contains only two "very preferred" items? I think that this decision will strongly increase the average number of "very preferred" features in future applications.