Self-study course · single skill · 5 weeks · part 2 of 2

Sending Morse

Timing is the heart of the code, and control of it rests entirely with the sending operator. This course is about the silences.

Progress0%

This bar is the word SEND, drawn to scale in real Morse timing. Tap it to hear where you are.

What you will be able to do

Send the alphabet on a straight key so a machine decoder reads it back cleanly, with your dah/dit ratio and all three gap classes inside tolerance — a stricter test than a human ear applies. And signal by torch at a flash rate inside the published safety threshold.

This course has a prerequisite, and it is not optional.

You need to recognise characters by ear without hesitation before you touch a key. That is Pierpont's position and his reasoning is explicit: most modern teachers hold back timing work until the student knows the characters by hearing, so that timing never becomes a way of analysing a character into parts. Sourced fact S1 In practice that means Copying Morse through Week 6 — sixteen letters, solid. Everything here works with sixteen; the last ten slot in as you acquire them.

Where this starts you

You can copy at least sixteen letters by ear at 12 WPM character speed. You have never sent anything. You do not own a key — the on-screen one measures your timing more precisely than a real key plus your own ears could, which is the point.

What this is not

Time budget

Roughly 11–12 hours across five weeks, published per block. Shorter than its companion because it carries two threshold ideas rather than three, and because you arrive already knowing what the characters sound like. Sourced fact S5

The two ideas this course turns on

How claims are marked

Morse instruction has an unusual evidence problem: the primary research is a 1935 German engineering dissertation and a handful of pre-1960 psychology papers, while nearly all modern writing about it is published by people selling an app. Every load-bearing claim below carries a marker so you can tell which is which.

Sourced fact S# traceable to a numbered source Inference my reasoning, in no source Vendor framing authored by someone selling the thing Contested the field genuinely disagrees Single study real but unreplicated

Sound settings

These persist. Leave character speed at 12 unless a block tells you otherwise — that number is the whole argument of Week 2.

Week 1~2.5 hrs · 2 blocks

Take care of the spaces

This week producesThe first characters you have ever produced rather than received — and a measured reason to care about silence more than sound.

You already know what these characters sound like. That is most of the work. What you do not yet have is a hand that can reproduce the rhythm, and an accurate picture of what your hand is actually doing.

Where you are expected to struggleYour first attempts will sound fine to you and decode as nonsense. That gap between how it feels and how it measures is the second threshold of this course, arriving early and uninvited.
Your first fist~1.5 hrs

Sending starts here and not earlier, on Pierpont's advice: most modern teachers hold back explicit timing work until the student knows the characters by ear well enough to recognise them without hesitation, so that timing never becomes a way of analysing a character into parts. You have sixteen letters at that standard. Now you can make them. Sourced fact S1

Not everyone agrees. Some teaching clubs pair every new character with "hear it, say it, send it" from the first lesson. The disagreement is real and I have followed Pierpont because his reasoning about why is explicit where the alternative is mostly practice. Contested Inference

Hold the pad. A short press is a dit, a long one a dah. Do not think about milliseconds — imitate the rhythm you already know.

Send — the spacing law~1 hr

Here is the finding that should govern everything you do with a key.

Pierpont, summarising the perceptual research: our judgment of the duration of brief sounds is poor, but we judge the relative length of brief silences much more accurately. If the dahs are substantially longer than the dits, the ear is satisfied and stops caring about the exact ratio. The silences, it does not forgive. Sourced fact S1

Which gives the oldest maxim in telegraphy, quoted by Pierpont as already traditional in his sources: take care of the spaces and the marks take care of themselves. Sourced fact S1

What this means at the key.

  • A dah that is 2.5 units instead of 3 is fine. Nobody will notice.
  • An inter-character gap of 1.5 units instead of 3 is catastrophic. Your letters merge and the receiving operator gets nonsense.
  • So: slow down the gaps, not the elements. Exactly the same principle as Farnsworth timing in Week 2, now on the sending side.

Send the word below and let the analyser show you where your spaces actually landed.

Week 2~2.5 hrs · 1 blocks

Your fist, measured

This week producesA numeric profile of your own sending — weight, three gap classes, consistency — and a diagnosis of your single worst fault.
Send — measured~1.5 hrs

Every operator develops a fist — small habitual deviations in timing that identify them the way a voice does. Pierpont documents this at length, and is clear that it cuts both ways: mild individuality is harmless and universal, but deliberately cultivated peculiarity crosses into unintelligibility, and some operators never notice they have crossed. Sourced fact S1

Send free text below. The analyser reports your weight, your three gap classes, and your consistency. Send at least a dozen characters or the numbers mean nothing.

The analyser is harsher than a human.

Pierpont's warning applies directly to this tool: machine decoders break down on poor sending, while the human ear and mind copy rotten code far better than any machine, compensating automatically for defects that would look glaring on paper tape. So a failed decode here does not mean nobody could read you. Treat the numbers as a stricter-than-life instrument — useful precisely because it will not do you the courtesy of guessing. Sourced fact S1

Diagnose~40 min

Each of these describes a symptom. Pick which fix applies and say why. Executing a named technique is not the skill; choosing among them is. Sourced fact S5

1. The analyser decodes your sending as one long nonsense character, and reports no inter-character gaps at all.

Your letter gaps have fallen below roughly two dit-lengths, so they are being read as the one-unit gaps that sit inside a character — everything has merged. This is the fault that ruins copy, and it is invisible from the inside because you know where your letters end. Fix: deliberately count three beats of silence after each letter. It will feel absurdly slow and it will not sound slow to anyone else.

2. Weight measures 2.4 and inter-character gaps measure 3.1. You want to be more readable.

Change nothing. A 2.4 ratio is inside what the ear tolerates — dahs are still plainly longer than dits, which is all the ear asks — and your gaps are correct. Chasing the ratio to exactly 3.0 buys you nothing a receiving operator would notice. This is the case where the numbers look imperfect and the sending is fine.

3. Your dit consistency is 0.40 and everything else is in tolerance.

Your dits vary wildly in length, which means the analyser's estimate of your dit — and therefore every ratio it reports — is built on a shaky base. Consistency is upstream of all the other numbers. Fix by slowing down until the elements are even; speed built on inconsistent elements is speed you will have to unlearn.
Week 3~2.5 hrs · 1 blocks

The gap that collapses

This week producesWords sent at the key rather than isolated characters, with the inter-character gap holding under pace.
Send — words~1.5 hrs

Single characters are the easy case. Words are where inter-character spacing fails, because the hand wants to keep moving.

Send each target. Watch the inter-character gap number specifically — it is the one that collapses under pace.

A physical cautionThe rapid pumping motion a straight key needs to send a run of dits carries a real repetitive-strain risk — historically common enough among telegraphers to earn its own name, "glass arm." Sourced fact S4 Keep sessions to the half hour this course already recommends, keep the wrist loose and the motion from the forearm rather than the fingers, and stop if anything aches. Speed is not worth an injury, and this course has no way to watch you for one.
Retrieve~20 min

Delayed, not adjacent. Spacing effects appear on delayed tests and not immediate ones, so the useful version of this is the annoying version. Sourced fact S5

1. Without scrolling: how many dit-lengths are a gap inside a letter, between letters, and between words?

One, three and seven. A dah is three, a dit is one. Those five numbers are the entire timing system. Sourced fact S2

2. Why does the perceptual research mean that a sloppy dah matters less than a sloppy gap?

Because human judgment of the duration of brief sounds is poor, while judgment of the relative length of brief silences is comparatively accurate. So the ear absorbs a mistimed dah and cannot absorb a mistimed gap. Every piece of sending advice in this course is a consequence of that asymmetry.
Week 4~2.5 hrs · 2 blocks

Sentences, and sending by light

This week producesA full pangram sent cleanly, and the same hand motion moved to a torch.
Where you are expected to struggleWord spacing is the last gap class and the one nearly everyone drops. Seven units is much longer than it feels — roughly a beat and a half of doing nothing, which your hand will refuse to do.
Send — a sentence~1 hr

Word spacing is the last gap class and the one beginners drop entirely. Seven units is much longer than it feels — roughly a beat and a half of doing nothing.

That sentence is the standard pangram and contains every letter, which makes it a complete sending self-test. If the analyser decodes it cleanly, your fist is readable.

Send with a torch~1 hr

A phone screen at full brightness is a serviceable signal lamp at night. Lamp mode below fills the screen with white on key-down — hold it toward whoever is receiving and send with the same hand motion you have been practising.

Full-screen is above the size exemption, so it is opt-in and speed-limited, and it stops the moment you release. Sourced fact S3

A real torch has no analyser attached, which is the point of doing it here first. Get the spacing measured indoors, then take the habit outside.

Week 5~2 hrs · 1 blocks

On your own

This week producesA cold sending assessment, and a written regimen you can run without opening this page again.
Assess — sending~45 min

Send the pangram cold, once, without practising it first. Clean decode plus gap ratios inside tolerance is the sending capability stated at the top of this page.

Write the regimen~45 min

One page, your own words, your own numbers:

  • How long is a sending session, and how often? Keep it short — the strain risk is real and the returns fall off. Sourced fact S4
  • What is your target sending speed, and which of the three gap classes is your weakest?
  • What are your current numbers for weight, each gap class, and consistency — and which single one are you working on?
  • How will you check your fist when there is no analyser? (Recording yourself and copying it back a day later is the standard answer.)
  • When do you stop optimising numbers and start sending to a human?

If you can write that page without consulting this one, the course is finished.

Last one. Your analyser numbers are all in tolerance, but an experienced operator tells you your sending is hard to copy. Who is right?

Probably the operator, and the reason is instructive. The analyser measures averages; it does not catch a distortion that appears only on particular characters — the dragged first dah in C, the stretched second dah in Q, the jerk in H or P. Pierpont catalogues exactly these as "swings," habitual peculiarities that pass unnoticed by the sender and make life hard for everyone else. Sourced fact S1 An average can sit perfectly in tolerance while one letter is consistently wrong. This is the limit of the tool you have been using, and the point at which a human ear becomes the better instrument.
Sources

What this rests on

Every load-bearing claim is keyed to one of these, and each was opened and read rather than cited from a summary.

  1. Pierpont, W. G. (N0HFF). The Art and Skill of Radio Telegraphy, 2nd revised ed. — particularly Ch. 9 (Sending and the Straight Key) and Ch. 15 (Timing). Tier 2 — canonical text, read directly. Freely distributed at the author's request. Source of the spacing-over-duration finding, the "take care of the spaces" maxim, the account of individual fists and "swings," the warning that machine decoders are harsher than the human ear, and the advice to defer explicit timing work until characters are known by ear. Note this is a practitioner synthesis rather than primary research: Pierpont collates decades of operator experience and earlier literature, and where he reports perceptual findings he does not always cite the underlying studies.
  2. ARRL / Bloom, J. (KE3Z). "A Standard for Morse Timing Using the Farnsworth Technique," QEX; with ITU-R Recommendation M.1677 for the element ratios. Tier 2 — canonical standard. Source of the PARIS 50-unit definition, the 1200/WPM dit, and the Farnsworth split. The arithmetic in Week 2 was verified by working the 31/19 decomposition out independently rather than taking it on authority.
  3. W3C. WCAG 2.1/2.2 Success Criterion 2.3.1, Three Flashes or Below Threshold, and the associated general flash threshold definition. Tier 2 — published standard, read directly. Source of the three-flashes-per-second limit and the small-area exemption (no more than 25% of any 10-degree visual field). The per-speed flash rates in Week 11 were calculated from the dit lengths, not taken from a source.
  4. Repetitive strain in straight-key sending — the historical condition known as telegrapher's paralysis or "glass arm." Tier 4 — general reference only. Included as a safety caution rather than a load-bearing claim. The condition is well attested historically; I have not gone to the clinical literature, so treat the mechanism as folk knowledge and the advice as ordinary ergonomic caution.
  5. Hattie, J. & Donoghue, G. (2021); Dunlosky, J. et al. (2013). Meta-analyses of learning techniques. Tier 3 — meta-analysis, appropriate here. Distributed practice and retrieval practice as the two techniques that survive aggregation; spacing effects appearing on delayed rather than immediate tests; interleaving forcing method selection.

The weak spot in this course. Its central source is a practitioner synthesis, not primary research. Pierpont collates decades of operator experience and where he reports perceptual findings he does not always cite the underlying studies — so the spacing-over-duration asymmetry, which everything here rests on, is reported rather than demonstrated in a paper I can hand you. It is consistent with what is known about auditory duration judgment, and it matches every operator account I found, and that is not the same as a citation. Inference

And the analyser is an instrument, not a judge. It is deliberately stricter than a human ear, which makes it useful for finding faults and unreliable as a verdict on whether you are readable. Pierpont's own warning applies to it directly. Sourced fact S1