So I decided to see what a public domain search at google books would turn up. The first search result was interesting, but when I got to Sir Richard Burton's Ultima Thule: or, a summer in Iceland, I paused. Sir Burton connects the Runic cipher to an Arabic cipher called "El-Mushajjar".
Soon I find an article Sir Burton wrote about "The Ogham-Runes and El-Mushajjar". In this article Sir Burton traces everything with a stick, branch, or stave back to Iran. The article is two parts genius and one part crackpot.
While reading related materials in google books I saw that later writers state that the Arabs based their El-Mushajjar cipher on the Norse cipher after coming into contact with Varangians. This explanation is simple and rational, but none of the sources put all the pieces together with simple diagrams in a step-by-step fashion.
So, that's what you're about to see...
The first to use this cipher were the Germanic tribes. The marks were used to refer to Runes. The Runes were separated into three groups of eight Runes. These groups were called Aettir meaning "eights". The sense of the word Aett is "group" or "family". The first group or family of Runes is called Freyr's Aett after the god Freyr. Frey'r's name starts with the first Rune of the Aett, Fehu (F). The other Aetts are called Hagal's Aett and Tyr's Aett and start with Hagalaz (H) and Tyr (T) respectively.
The first two types of cipher Runes described by Saint Gallen are called "iisruna" and "lagoruna". These are the Runes Isa and Laguz. In Saint Gallen's manuscript, the short Runes represent the Aettir. One short Isa or Laguz represents the first Aett, Freyr's Aett. Two short Runes represent Hagal's Aett. Three short Runes represent Tyr's Aett. Directly after the short Runes are taller Runes. The number of taller Runes tells you the index of the Rune.
The third type of cipher Rune described by Saint Gallen is called "hahalruna" and is based on the same pattern as the others. We make a vertical line. To the left of this line we make one, two, or three marks based on which Aett the Rune is in. Then we make a number of marks to the right of the center line equal to the Rune's index. If you slant the lines upward you get the forms you see in my table above. No matter which way the lines point, they mark either the Aett or index. Sometimes you see the lines facing upward on one side of the stave and downward on the other side.
The fourth type of Runic cipher described by Saint Gallen is called "soofruna". It's the same as the others, but with dots.
In all four examples, Saint Gallen has written the name CORVI.
As times went on, the Runes underwent some changes and were reduced in number to sixteen. The following table shows how the old cipher was applied to the Younger Futhark.
I could not find a good free Younger Futhark font, so I just wrote it out quickly on graph paper. I attempted to copy the rune forms found in Johan Gustaf Liljegren's Run-Lara.
As you can see, tree runes were used by the Norsemen before they journeyed into Arab lands and were still used in the Viking age.
It was the opinion of Sir Richard Burton that the tree runes were an Arab invention that was borrowed by Varangians and adapted to use with their own runes. Why it did not occur to him that the transmission was from northern runes to Arabic is uncertain.
The oldest Arabic manuscript I could find that contains the cipher called El-Mushajjar is that of Ibn Wahshiyya (9th century):
The alphabet of Dioscorides the philosopher, commonly called the Tree alphabet. He wrote on trees, shrubs, and herbs, and of their secret, useful, and noxious qualities in this alphabet, used since in their books by different philosophers.
Sir Richard Burton suggests that it was an Arabic translation of a work by Dioscorides that was written in this cipher and then attributed to Dioscorides. Let it be written down that I officially agree with Sir Burton about this one issue.
In his article in "Transactions of the Royal Society of Literature of the United Kingdom", Sir Burton gives the following form of the cipher:
The Arabic alphabet was originally arranged in a different order than you see it today. The letters, like the letters of so many other alphabets were also used as numbers. The change to Arabic numerals (which the Arabs call Indian numerals ;) had already occurred by the time Ibn Wahshiyya recorded the cipher. This can be seen by looking at many of the other ciphers in Ibn Wahshiyya's work. This leads me to believe that the ciphers recorded in Abjad sequence were already in use by Arabs before the new letter sequence arose and those with the new sequence were developed closer to the time Ibn Wahshiyya recorded them.
Of this system Idries Shah (The Sufis, 1964) writes:
While the Arabic letters have equivalents up to one thousand, the Hebrew alphabet has equivalents only up to four hundred, inclusive. For mnemonic purposes this arrangement of letters is always memorized as follows, as a string of meaningless words, adding diacritical points to make pronunciation possible:
ABJAD HAWAZ HUTY KALMAN SAFAS QURSHAT THAKHDZ DHATZAGH.
In Persian, Urdu and other non-Semitic languages, the letters are given slightly different sounds in some cases, but this does not affect the use of the letters, whose numerical values remain constant.
Date names, dates of birth or death, words expressing the character or aspirations of a person, all are often evolved from the scheme. Ignorant repetition of the meaningless words has in some places endowed the barbarous Abjad "words" with spurious baraka, the belief in special inherent functions, but this belongs to the realm of repetitious magical procedures and is not important.
It is at this point I would like to point something out. The Abjad system is based on number. So was the Elder Futhark. When viewed as a base-8 tally system, Hahalruna not only have a one to one relationship with the position of the Runes of the Elder Futhark within the three Aetts, but also give the numeric values of the Runes. This is not true for the Younger Futhark or Abjad adaptations. It is still a fine cipher system no matter what alphabet it refers to.
Here is a little table I made up showing the way in which the cipher was applied to Arabic:
So, yeah, this is what I did when Wikipedia went down for a day. What would I do if the internet kill switch were used?
Hahalruna and El-Mushajjar by Jason Bales is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.
Based on a work at greybookofrunes.blogspot.com.
I was up all night reading the internet again and just came across this text at Princeton:
it's a text titled "Treatise on ancient, alchemical and magical alphabets"
I managed to find El-Mushajjar written in the left margin on this page:
As marginalia goes, it looks pretty clumsy. Notice the modern letter order written underneath the cipher and then crossed out. The Abjad order is written above.
Take a look at this 18th century manuscript:
1:10 , 3:9 , 2:9 , 1:9 , 3:8 , 2:8 , 2(3):8 , 3:7
The second character, reading from the right is smudged and looks to me like the transcriber originally had 3:8 , but scraped away parchment to make it 2:8. As you can tell by comparing the tree runes in this manuscript with those in my table above, there's something different. Whereas there are only eight abjad words, the final tree rune implies at least ten.
The solution to this problem is simple. The 18th century version of this cipher makes its divisions based on the hijā’ī order of modern Arabic.
More on coded Runes: