The 1989 Women's World Championship Teams (contesting the Venice Cup) in Perth; Australia, the USA vs. the Netherlands. On the deal below, something goes seriously wrong on the Dutch side.
But then an American player glitches and saves the Dutch women:

E/NS		♠	—
		♥	A 7
		♦	K 5 2
		♣	A K Q 8 6 5 4 3
♠	10 9 4			♠	K 8 7 6 5 2
♥	K 9 8 6 5 4			♥	J
♦	8 4			♦	Q J 10 6
♣	9 2			♣	10 7
		♠	A Q J 3
		♥	Q 10 3 2
		♦	A 9 7 3
		♣	J

¹ Weak two: six-card suit, 6-10 HCP
² Meant as control-showing; interpreted by South as 'two suits'
³ Meant to play; interpreted by South as 'choose between the minors' (consistent in view of ²)

Bakker's 4♠ bid led to a major misunderstanding between North and South. Bridgevaria doesn't know which explanation was correct according to NS's bidding system, but does wonder what purpose North's 4♠ bid had, if she intended to play 7♣ all along. After all, no (non-jump) bid from South over 4♠, like her actual 5♥ bid, would give North any extra information. So a direct 7♣ bid over West's 3♠ bid would have had no disadvantages compared to North's actual bidding, while it had one huge advantage: it would have been unequivocal. 
North was unhappy (to say the least) with the developments but couldn't but pass 7♦. After all, she didn't have the ♠A, so bidding 7NT would be ridiculous. So she just hoped South had a good four-card (or five-card) suit: ♦AQJx(x).

The Dutch pair was thus heading for disaster, but behold! In view of her impressive holding in trumps Gwozdzinsky, East for the USA, felt certain she could beat 7♦. She expressed that feeling by doubling.
Gielkens trusted the double. Partly perhaps because she wasn't sure about the bidding so far. Furthermore, she may have feared East to ruff the opening lead. So Gielkens bid 7NT.
This time West doubled. So far East would not have worried too much, fully expecting 7NT to go down as well. But then dummy went down, unveiling a solid eight-card suit in clubs!
West did very well by leading the ♦8, one of the few cards that didn't give South the contract.
Declarer won in dummy with the ♦K and cashed the ♥A and all of the clubs, thus squeezing East in spades and diamonds. This was the situation with the last club to go:

E/NS		♠	—
		♥	7
		♦	5 2
		♣	3
♠	10			♠	K 8
♥	K 9			♥	—
♦	4			♦	Q J
♣	—			♣	—
		♠	A Q
		♥	—
		♦	A 9
		♣	—

Declarer knew from the bidding that East had started with six spades (and at least four diamonds), so all she had to do was observe how many spades East discarded. So in the layout shown declarer knew East had kept two spades.
On the ♣3 East was in trouble. If she were to throw a diamond, South would discard the ♠Q and make the last three tricks with the ♦A9 and the ♠A.
If East were to discard a spade, declarer would know that East had bared the ♠K. So South would discard the ♦9, making the last three tricks with the ♦A and the ♠AQ.

Moral of the story: if the opponents bid a grand slam, which you can beat because of your trump holding, be very quiet. If you double, the gain is usually negligible, while you lose heavily if the opponents escape to another, makable grand slam.
We will take a closer look at this. At the other table the American NS bid 6♣ and made an overtrick: +1390 to the USA.
Now if:
- 'Our' East passes 7♦, South will suffer a two trick defeat, +200 to the USA. Team USA thus wins 1590 points, 17 IMPs.
- 'Our' East doubles, as she did, but NS do not escape to 7NT. Against 7♦ doubled EW score +500 now. Team USA wins 1890 points, 18 IMP. Meaning the double results in only a one IMP extra profit!

In practice, the double resulted in 15 IMPs loss, against a 17 IMP profit if East had passed. So the double actually lost 32 IMPs.
The double therefore either gains 1 IMP or loses 32 IMPs. The conclusion must be that this double was not a good bet, to say the least.

PS: Another reason not to double can be that the double warns declarer about the bad distribution. Sometimes it will enable him to find a winning line of play, which he would not have found without the double. Of course this is only the case if the contract would only go down one or two. The double wouldn't be very profitable anyway then − in IMP-scoring that is.

PPS: Want to read more about doubles like East's? Then read: 'Penalty double or pass', three consecutive columns by Peter van der Linden. Click here for part one, follow the links in the article to the other parts.