'Regarding binary system, I meant when considering only individual letters' Why would I consider only individual letters? You can also say Morse Code is ternary considering only letters AND inter-character pause. I think I need to consider the system as a whole.

Ok thank you for your response. Although I must say it is still not clear to me. I still think Morse Code is ternary, because I can easily say 102102210 where I just said: AA A. I needed 3 different types of information piece. Anyway maybe it is me who is having trouble understanding, thanks for the help. I will probably have an "a-ha" moment tomorrow or so..

@KorayTugay I see now what you mean with this example. Yes, you have a very good point. You can consider it ternary that way. I do not thing the exact duration of the short and medium gap is too important, and you may consider one as twice the other (approximately). I hope my words did not offend you, as I had not realized you might take it that way. I am upset when wikipedia contains silly stuff, which occurs too often, and misleads a lot of people. And my criticism was adressed at wikipedia.

For some applications, there is one good reason for the quinary (Wikipedia) approach over quaternary: The quinary system can be statelessly decoded into audible morse code while the quaternary system can not (as it requires: if current symbol is dot/dash and previous symbol was dot/dash then insert signal separator rest). You could decode quaternary statelessly if you inserted a dot-length rest after every symbol and reduced the letter separator to 1 rest and word to 5 rests, except M.1677 does not explicitly prohibit consecutive word spaces, which breaks that.

Iagreewithyouthattalkingaboutgapsbetweencodesareonlyofinteresttoinformationtheor‌​‌​ists.Realhumansdon'tneeduselessgapstounderstandEnglish.It'snonsensethatStackexc‌​ha‌​ngecommentlimitsincludesthisgapincommentlengthcharactercount.Thatiswhyfromnow‌​Iwil‌​lneveragainwritewithgaps.

The point is, the inter symbol gap is important part of the code, it carries important information to allow you to distinguish between three dots and a dash. Not taking into account of the gap would allow you to encode information denser than information theory would otherwise allows, and that is what makes no sense.

@LieRyan I removed my previous comment and explains things precisely in a more technical top addendum to my answer. The point is that while the inter-element gap is necessary to encode one element (such as a dot or a dash), at the level of dots and dashes it has no more meaning than the horizontal bar in the letter T. The alleged quinary encoding was mixing dots, dashes and inter-element gap. That was nonsense. But look at the more formal presentation I give. Note, the ternary encoding I suggest is not that of Jason C.

@babou Question: There are two strong argument for quinary: 1) The signal gaps are implicit in ternary (due to notation) but explicit in quinary, the ternary representation does not contain the "signal gap" primitive of the standard (Lie Ryan also mentioned this above), and 2) the "sep" symbol in ternary requires context for its semantics, with quinary (and quaternary) it does not. All ternary choices are essentially a compressed coding, and signal separators are actually the symbol separators used in the encoding itself, which are "hidden" here. Thoughts?

(That is, the strongest argument for quinary is every symbol is meaningful without context, it relies on nothing implicit in the coding itself, and represents the 5 basic primitives defined in at least International Morse Code.)

@JasonC Did you read my modified answer? My ternary is not your ternary, and it three symbols are binary encoded. I am using only prefix coding, and have no problem of interpretation at any time. The system remains extremely simple. Prefix coding is the key. This is the only structured, simple and compelling view I have seen. BTW, 57-ary is completely misguided, and has nothing to do with the topic at hand.

@KorayTugay Since you set a bounty, you are not satisfied with answers. So it is fair for me to ask what you see missing, incorrect, incomplete or unconvincing with mine. There is clearly a binary component, since the stream is a bit stream. Thenthe basic constituent are composed of 3 symbols coded in binary, And the letters and interword spaces are encode in these 3 symbols. And all is with prefix codes allowing easy and unambiguous decoding.

@KorayTugay So you may wonder why I did not come up with my current answer the first time. The reason is that I was missing a key information. I did not know the standard specified that everything was measured precisely in units of time, which implies that it starts from a binary stream. Since Morse is also done by hand, I thought the basic signal to be with approximative durations, which precluded seing it as a binary stream. This naturally lead to a different line of reasonning, which I would still be defending, if I did not know now that it uses a binary stream communication channel.

@KorayTugay Yes, binary. According to the standard, the information is transmitted as a stream of bits (which I did not know before). Hence you necessarily have a binary representation of the message transmitted, but unrelated to dit and dah. This could be used to send any kind of binary representation as found in computers, such as integer in binary numerals. But, logical analysis of Morse code shows that it is used to encode simply 3 symbols called dit, dah and sep (I made up the last name, for inter-letter separation), using a prefix code that is very easy to encode and decode.

@KorayTugay This set of tree symbols is the basis of what is seen by the public as the Morse code. That is the public knows only of dit and dah to make letters, and sep to separate them. Not knowing of the binary level, all he sees is a ternary stream, composed with these symbols. But what is needed is characters for writing English and numerals with punctuation, and a space between words. So these characters are encoded as a sequence of these three symbols, as indicated in my answer, again using a prefix code, efficient and unambiguous. Since it uses three symbols, it is a ternary code.

@babou I am too tired right now, I will check this thread tomorrow. But please see my latest edit in my question.

Please move any further conversation to chat.stackexchange.com/rooms/21638/…