Black just played the marked stone... What's whites best attempt at saving his marked group? (Enjoy!;) Reuven
Tderz: Possible starting moves which come up to mind are:
White a, b, c, d and e.
Hence, Whites starts: I see that a and b, are miai, the black corner has more fighting liberties than the 5 (now 4) white dame. Continuing with a looks more promising:
(before screaming, look below -)
It looks like a miscalculation for Wc, but could be used as large scale attack on the black center group:
yet White can continue with or c and separate black's group.
Not a success for Black, got only some white stones , but center group is cut-off and weak; the lower edge has a bad endgame.
I think ,a is now futile because of b;
and starting with b does not work because of Bc, Wa, Bd and White has damezumari problems.
White e and f are also quite unrelated to the semeai below in this diagram.
at present dia. 1-3 look best, but comprise an error: is answered by - and White has liberty problems. Hence, one could try and leave out the liberty losing exchange - ind. 
Now White considers playing h or b in order to destroy the eye-shape of the black center group, but that makes no sense as Black can connect at a or play i (if White a). Hence , now coming back safely to White g and capturing the is good enough.
After Black must find the right moment to play a or i or perhaps first in the center around m. Black a is big and honte and allows free play afterwards. (Hope I didn't make stupid mistakes here)
Black ends in sente and could make later a good endgame around x to z for capturing .
It is sente because White a, black r, white i does not attack much Black's center group (having the ko at k, an 1/2 eye at c and space around m).
Is this compromise the most likely outcome?
If now White is squeezed to , then Black is connected on top
but the are lost after (not a, because Black would lose 2 stones after white b and the ko at c would only help White).
Already seeing this shape -- tells us that White played inefficiently here.
White should look for better options.
Yet, similar to dias 5-6, White could try again to ,
but again would fail after white miai of a and b.
White could compromise with black via to .
This is a likely outcome. Both White and Black have connected their groups.
(if Wa, then Bb) Black will lose territory in the endgame around the x-points
In dia  I state something about the corner having more fighting liberties, but after 2nd thought, it does not seem so straightforward: After , , Black must prohibit all connecting options to the left and takes away a liberty with .
The have 4 liberties left, technically the corner can be killed easily (made nakade, common shape) by but in exactly how many moves?
Black has a 4-nakade with (5-3=2) 2 liberties + the ko evolving at a + sente. If we deduct 1 liberty for the sente from White's 4 liberties, we guess it's 3 vs. 3 liberties in ko.
... after some ko threat by White
but beware of the later ko at d, hence perhaps e see dia  ...
Black neglects the ko threat on the center group, because cutting is Gote for White and Black can start a new ko with . White will have lost two fights and then perhaps the game.
That's shocking news unkx80 :-) Tderz I should absorb your columns better.
It's a two move approach ko for White. But two questions remain in my mind:
starts a ko on the White group on a large scale. However, White seems to have a lot of internal threats. Still I think that Black may expect more of this than from the two move approach ko mentioned above.
Tderz: reply to Dieter's question 2: ko? First, it seems to me now that Dieter's sagari is better technique (e.g. having a liberty more at a) than hane+connect, but I will check on this.
The ko does not seem so effective, because the have lost liberties and are therefore a source of many ko threats - if not more !:
After Black hardly wants to sacrifice the Black stone below .
Besides that any of the previous white moves - could/should have been used for ko at , a 2nd ko evolved, hence looking like double-ko.
Black can play c or d, I think it's similar.
... afterwards, astonishingly a and b seem to lead to the same (German "Zugumkehrung" = move inversion?) good result for white - notice that this seems better than just double ko. In first instance, I preferred a.
White could exchange for for separating (Black replies, after all it's big), then attack at . Black would not answer White a at b (the 3 stones are useless now), rather escape making eye-shape around = or further.
Are we ready for conclusion?
Is unkx's dia , the simple sagari-tesuji? making miai of killing the corner or connection the simple & best answer to this problem?
Rich: Ummm... doesn't that diagram show white one liberty short of survival?
Compromise dia  leads to a bad endgame on the lower side for black. Knowing that white in all variations seems able to connect somehow - and that also the connection of the black center group is at stake - Black should try to get sente and/or a good endgame on the lower side.
White plays here to avoid dia . Black finds this result un-attractive, and might even prefer diagrams  to , where Black loses 7 (about 16 points, but white could lose and black keeps the lower edge)), but in sente!
So, is dia  best for Black and therefore the fairest, most likely compromise?