Morse Code vs Tap Code: Two Systems, One Wall of Constraints

They are constantly confused for each other — prisoners tapping, someone blinking, a movie hero rapping on a pipe — but morse code and tap code are two different engineering solutions to the same problem: sending an alphabet over a channel that can barely carry anything. One encodes letters in the *duration* of two elements. The other encodes them in *counts* of a single element. Everything else — speed, robustness, learning curve, the historical moments each dominated — falls out of that one design choice.

This page compares the systems properly: the encoding math, the channel behavior, the learning economics, and the use cases where each is still the right answer in 2026. It draws on the historical record covered in the tap code history and the military stories, but the analysis here stands alone. Both systems are interactive on this site — the morse chart with audio playback for one, the tap code tool for the other — so every claim below can be verified by hand in a few minutes.

How morse encodes a letter: duration

Morse uses two elements distinguished by time: a dot, one unit long, and a dash, three units, with precise silences between them — one unit between elements, three between letters, seven between words. Letters take one to five elements, and the assignment is frequency-optimized: the most common letters in English get the shortest patterns. E is . and T is - — one element each — while rare letters pay full price (Q is --.-, four elements). A word like HELP is four letters of very different weight: H .... E . L .-.. P .--. — eleven elements total, with the single dot of E arriving almost for free. Averaged over ordinary English text, the code lands at just under three elements per letter.

The system's genius is what the frequency design implies: the sender's effort tracks the receiver's expectations. Common words flow fast, rare words slow down, and the total channel time per sentence is close to the theoretical minimum for a two-element alphabet — a property engineers would rediscover a century later as Huffman coding, for which morse is the famous ancestor. The price of that efficiency is cognitive: the receiver must judge *duration* accurately, at speed, because a long-seeming dot is a dash and there is no other punctuation. Morse is fast because it trusts timing; that trust is also its only real vulnerability.

How tap code encodes a letter: position

The tap code makes the opposite bet: abandon duration entirely. Write the alphabet into a five-by-five square (C and K sharing a cell in the modern standard), and address each letter by two counts — its row and its column. The letter T, in row 4, column 4, is four knocks, a pause, four knocks. The letter A is one and one. There is exactly one element type — the knock — and exactly one skill: counting. Duration is irrelevant; a slow knock and a hard knock are the same knock, and the grid does the rest.

The costs are as symmetric as the design. Every letter costs two number-groups, and because the grid is square, every letter costs the *same*: an E and a Z and a Q are identical in effort, six knocks on average (rows and columns average three each), so the code cannot exploit English letter frequency at all. Where morse spends just under three elements per letter, tap code spends about six knocks plus two pauses — more than double the channel events, with no discount for common letters. It is a fixed-length code: maximally fair, maximally simple, and maximally slow.

The math: nine elements versus twenty-one knocks

One worked example settles the comparison. Take the most famous distress signal in history, SOS. In morse it is three letters of three elements each — ...---... — nine timed elements, sent as one fused run in about two seconds at a modest operating speed. In tap code the same message is S (row 4, column 3: seven knocks), O (row 3, column 4: seven knocks), S again (seven knocks): twenty-one knocks plus the pauses that separate groups. Same message, same wall of constraints — and the timing-based system finishes roughly four times sooner in wall-clock terms, even at a slow, deliberate sending speed.

The information-theory view explains why the gap is structural. Each tap-code letter carries a fixed log₂(25) ≈ 4.6 bits — always, because every letter uses two numbers from one to five. Morse's variable-length letters average about the same information per letter of English text (Shannon's estimate of English letter entropy is in the same range) while spending fewer channel events to deliver it, because the short codes are assigned to the high-probability letters. Fixed-length codes buy simplicity and uniform error behavior; variable-length codes buy speed. The prisoners and the radio operators each chose correctly for their channel, which is the real lesson of this comparison — the code is a function of the channel, and no code is best on all channels.

Which system survives a bad channel?

Now corrupt the channel, and watch the rankings invert. Send morse through a concrete wall, by exhausted fists, and the timing collapses: knocks arrive muffled or sharp depending on the wall, the sender's knuckles, and the receiver's distance, so dashes get read as double-dots and dots as clipping — the letter boundary, encoded only in silence length, disappears first. This is not speculation; it is the documented reason twentieth-century prisoners of war, many of them trained military signalmen who *could* send morse, built their cell networks on the counting square instead. The historical record from the German Stalags to the Hanoi Hilton is unanimous on this point, and it is the single strongest real-world A/B test of the two systems in existence: professionals, choosing freely, picked tap code for the wall and kept morse for channels they controlled — a blink, a cough, a fingertip on a tin cup.

Corrupt the channel the other way — make it fast and clean, an electrical tone on good wire or radio — and tap code has nothing to offer. Its fixed cost per letter caps throughput at a few letters per minute even between skilled partners; morse operators, released from channel brutality, run twenty to forty words per minute routinely and have exceeded seventy-five in documented competition. The engineering summary fits in one sentence each. Morse: maximum speed, demands timing fidelity. Tap code: minimal demands, pays in speed. For a modern corollary of the same logic in the audio domain — where timing survives but noise does not — see the degraded-signal practice built into the audio decoder.

Which is easier to learn?

The learning economics diverge even harder than the speed economics. The tap code can be taught in an afternoon: draw the square, explain row-then-column, drill for twenty minutes, and a pair of motivated people are spelling through a table leg the same evening. Total fluency — automatic counting, comfortable speed, error recovery — takes a few weeks of casual use. The grid never changes, there are no new symbols, and nothing needs to become reflexive beyond the alphabet's position in the square, which most people internalize by writing it out a few times. You can be functionally slow today and functionally adequate next week; our tapping game is built on exactly that curve.

Morse is the opposite bargain: trivially explained, genuinely demanding to acquire. Reading at conversational speed requires the letter patterns to become automatic — heard as sounds, not counted as elements — and that automation is a real skill acquisition with a well-documented plateau (the famous ten-words-per-minute wall where counting strategies fail and must be replaced by whole-pattern recognition). Done right, with modern full-speed methods, a motivated learner reaches functional copying in a few weeks of daily short sessions; done wrong, with charts and mnemonics, learners stall for years. The learning guide exists because the method matters that much. In exchange for the harder on-ramp, morse delivers a skill that scales — speed, lifetime fluency, and compatibility with the entire world of radio, which brings us to use cases.

The side-by-side comparison

Condensed to one view, here is how the two systems stack up on the dimensions that matter. The table is the whole argument of this page in nine rows; the paragraphs above are the evidence for each row.

Which should you learn in 2026?

Match the code to your actual use case, not to the romance. Learn the tap code if your interest is escape rooms, puzzle design, tactical or family signaling games, or plain curiosity about the prisoners' system — it is cheap, it is genuinely useful through walls and across tables, and our tap code page covers the full grid with an interactive encoder. Learn morse if you want anything that touches radio (the amateur license tracks are morse-free in most countries now, but the CW bands remain the hobby's deepest tradition), historical signaling, audio puzzle-solving, or the assistive use cases where a two-input keyboard changes lives — the Gboard accessibility story is the modern proof.

And if you are designing rather than learning — an escape room, a game level, a family scavenger hunt — choose by your players' channel. A repeating lamp or speaker they can see and hear cleanly? Morse, with a legend, and the message short (see the escape-room design guide for the format conventions). A knock through a wall where timing will smear? The square, always. The prisoners of three wars would approve of either choice; what they would not approve of is using the wrong one and blaming the code.

Try both in ten minutes

The comparison above is testable tonight, which is the best thing about it. Open the translator and key SOS at slow speed — watch the nine elements and feel the timing. Then open the tap code encoder and send the same message by counts — twenty-one knocks — and feel the difference in your hands: rhythm versus arithmetic. Ten minutes of doing both teaches more than any table, including this page's.

Then, if you want the full experience our two historical lineages offer: the Koch trainer for the morse ear, the quiz for letter recognition under time pressure, and the dot-dash primer for the element grammar underneath it all. Two systems, one wall of constraints, and a single lesson they teach together: no code is superior in the abstract — the channel chooses the champion, and the wise signalman knows both.