As long as the plaintext has some sort of recognizable structure, it wouldn't be that hard to identify where a letter has been omitted. Here's an example (generated using the default settings of this online Enigma simulator):
Plaintext: THEQ UICK BROW NFOX JUMP SOVE RTHE LAZY DOG
Ciphertext: OPCI LLAZ FXLQ TDNL GGLE KDIZ OKQK GXIE ZKD
Suppose the 15th letter of the ciphertext (
N) is omitted. This would result in the following decryption:
Ciphertext: OPCI LLAZ FXLQ TDLG GLEK DIZO KQKG XIEZ KD
Plaintext: THEQ UICK BROW NFVH JAOW FSNA TPPT ESUW BM
Clearly there's something wrong after the word
BROWN, so let's reset the machine to its initial settings and insert an additional
X after this word in the ciphertext:
Ciphertext: OPCI LLAZ FXLQ TXDL GGLE KDIZ OKQK GXIE ZKD
Plaintext: THEQ UICK BROW NHPX JUMP SOVE RTHE LAZY DOG
This is much better, but now we have the word
HPX, which makes no sense. A consecutive run of unlikely characters suggests that we inserted the extra character in the wrong place.
F in the original plaintext was probably correct, and the middle word of the plaintext must be
F.X (with the
. representing a character that was lost in transmission). There aren't many words that fit this pattern, and it isn't hard to guess which one is correct.
Obviously this method would be harder to use if a message contains clusters of missed and/or incorrect characters. But in that case the wireless operator at the receiving end could just request a retransmission of the message. I don't see why they would have to wait 24 hours.