Research
The research behind every Moon sign
Moon type has no living standards body and its official manual is not public. Every rule in this font therefore comes from a document that somebody can go and check. This page says which documents, what they settle, and where the font is guessing.
If you only read one thing here, read the next section. It is the rule the whole project rests on.
The rule
Three kinds of sign, never mixed
Attested
A document shows this sign, and the document is named. Most of the font is this.
Conjectured
Nobody has found a document. Somebody proposed a shape by following the pattern of the signs that are attested, and it is labelled as a proposal.
Extension
Modern print needs a character Moon never had. This project added one, following an existing Moon pattern, and says so.
The distinction is kept in the catalogue sign by sign. It exists so that a guess can never quietly become a fact — which is the way a research record decays.
The test
A real page, reproduced exactly
Rules taken from a chart can be read wrongly. So the font is tested against a page of real Moon: page 42 of The Moon Magazine of May 1927, photographed and transcribed by Michael Everson.
The font sets the same text and the result is compared with the transcription, word by word. All 179 words have to come out identical, every time the font is built. That test has changed the font more than once: it is how this project found that the 1927 magazine used a level of shortening between Grade 1 and Grade 2, which is now one of the switches.
179 words
The whole page, compared on every build.
0 mismatches
The only acceptable result. A rule change that moves this number is wrong, or it is a finding.
Built from sources
Nothing in the font is drawn by hand. Every outline comes from a rule, so a shape cannot drift.
The sources
What the rules come from
These are the documents that settle the most. They are cited here, not copied: several are still in copyright, and the ones that are not belong to the libraries that scanned them.
- Moon's own Light for the Blind, 1873 and 1877 — the alphabet, and specimens in eleven languages.
- The RNIB alphabet sheet — the modern letters, numbers, punctuation, and the list of short forms this font applies.
- Everson's 2011 proposal to encode Moon in Unicode — a repertoire, the historical figures reproduced in it, and the 1927 page used as the test.
- Rutherfurd, 1898 — the biography, and the publisher's own list of everything printed in Moon.
- RNIB, 2008 and 2009 — who reads Moon now, how large it is set, and what its problems are.
- The Dotted Moon project, 2003 — what happens when Moon is made of dots instead of lines, tested with twelve readers.
The full record — every source, when it was read, what it established and its rights — is in the sources file, and what is still being looked for is in the material tracker.
Still missing
Documents we would like to read
Each of these exists. None of them has been read by this project. If you hold a copy, or work somewhere that does, please write.
-
foundational
RNIB, Moon Transcribers Instruction Manual, April 1996
The normative Grade 2 rules, the complete exception list, and §3.4 — the capital indicator's shape
Likely held by RNIB Research Library, RNIB Archive
-
very high
The Moon System, Regency Press, Brighton, 1962 — complete
It already settled the numeral history and the eight base signs from two pages. The rest is likely to settle as much again
Likely held by RNIB; no copy located. Two pages survive as photographs in Fulas 2021, p. 230
-
very high
Moon, A Simplified Alphabet for the use of the Blind — an original sheet
Moon's own instructions at full resolution: the classified alphabet, the star numerals, and whether other editions of the sheet differ
Likely held by Library Company of Philadelphia; one photograph in Fulas 2021, p. 218
-
very high
Armenian Moon: Psalms 34 and 86; John, chap. 3; Primer — and Armeno-Turkish Matthew (2 vols)
The 26 Armenian letters Everson leaves blank
-
very high
Portuguese John, chap. 3
Whether Portuguese Moon marked accents, and whether it contracted anything
-
very high
Cryer, Home & Wilkins, User Requirements for Moon, RNIB, 2011
What teachers and specialists said Moon readers actually need
Likely held by RNIB. Known only from a later citation; request it with item 1
-
very high
The Moon Magazine, vol. 86 no. 5, May 1927 — the complete issue
Many pages of running text instead of one, to test the contraction rules against
Likely held by RNIB Archive, British Library, Perkins — and US libraries received it monthly into the 1940s
-
very high
Clean Duxbury English/Moon output, outside demo mode
Whether seven candidate contractions found in its tables are real rules
Likely held by Any licensed user, or Duxbury Systems; the corpus to run is in duxbury-mac-inspection.md
-
high
Possum Moon-Writer User's Handbook (draft, with tactile graphics)
Whether the ss05 letterforms and their stroke order are right
Likely held by The Dot Experience, formerly the APH Museum, accession 1992.359. A Perkins Archive photograph shows the machine
-
high
Campbell, The Gospel of Matthew Adapted for Blind Readers on the Island of Formosa, 1888
The complete Revised Moon alphabet and its seven tone marks
Likely held by University of Glasgow Special Collections; one page is reproduced in Chiu 2023
Four filled circles means the whole project is waiting on it; three means it would settle a lot. The full queue, what has already been read, and which searches came back empty are in the material tracker.
Honestly
What this project does not claim
- Nobody has read it by touch. There has been no test with a Moon reader and no test on a machine that presses paper. The sizes follow published advice, not measurement.
- Moon is not in Unicode. The font uses the Latin alphabet, and also the code points of a 2011 proposal that was never approved. The roadmap has since moved the space it would occupy.
- Some signs are guesses. Where no document could be found, the catalogue and the font manual say so, sign by sign.
- The official Grade 2 manual is not public. What is here comes from the documents that are public, and from the printed page the rules are tested against. Parts of the manual survive inside other people’s software; that is not the same as having read it.
- This is not certified transcription. A font sees letters, not meanings. For a document that has to be right, use dedicated transcription software and keep the source text.