Interesting thread. One thing I notice is that the disagreements about resemblances always rely on global impressions of a letter's shape as a whole. This is a weakness of discussions of relationships between scripts in general. I notice the same difficulty in the literature on the relationships of scripts in the Malay archipelago: different authors have different impressions of relationships but there is no principled way in the literature to evaluate one person's impressions against another's. As long as things stay this way, comparative graphonomics will not be a scientifically-based area of study. This is why I have worked out an analysis of structural features of letters that define natural classes, much like in phonology. This analysis lets me describe exactly how letters relate across scripts and has helped me show on a clear basis how the Malay archipelago scripts relate to each other and to continental Indian scripts.

I'm planning on doing something similar for these scripts as well. I agree that Armenian and Georgian scripts are most likely derived from some Persian descendants of Imperial Aramaic. Kharosthi (9) shows systematic relationships across letters to Arsacid Parthian scripts (4, 5) and (mirrored) Armenian shares a lot of structural features with Kharosthi (actually, more than this table lets on). This is surprising in geographical terms, and is a strong argument for an immediate common ancestor.

I finally traced this table down to Isaac Taylor's 1883 book The Alphabet: he points out the same similarities and proposes that Armenian is likely derived from an Arsacid forebear and Georgian from Sassanid. His discussion is worth reading; the book can be downloaded in various versions via Google Books and Archive.org. He actually has a lot more interesting things to say elsewhere, such as his discussion of the likely Safaitic origin of Brahmi. Although the similarities were not conclusive in his time, I think he was dismissed too easily in favour of an Aramaic hypothesis that has never been much more than mere conjecture. Further discoveries in North Arabian scripts since Taylor's time have brought to light a mass of evidence that make a combined Safaitic-Taymanitic-Thamudic origin much more likely.

I think you missed a few cognates: the Armenian counterpart of He is closely related to the Arsacid shapes in 4 and 5 (more so than to 6 in terms of how the strokes are made), of Waw is also closely related to the Kharosthi (9) shape, of Zayin to 5 and 9, of Teth to the original Aramaic shapes in 2 and 3, of Samekh to 4-7. These correspondences aren't so much in the overall shape as in the way the strokes that make up the letters curve or bend abruptly, touch or intersect in pairs to be compared.

It would take me forever to actually describe a theory of letter structure and how letters relate by changes to their underlying structure, so I'll just invite people to reflect on how the original classical Latin letters in the first column below relate to those in the following columns – not impressionistically and globally, but in terms of the way the strokes they are put together with related across related letters. (I placed Y before V because although it is a later borrowing from Greek, it is at the same time the original shape that V came from. It's also worthwhile thinking of the relationship of the original Greek shapes to the Latin ones and to the modern lower case Greek letters.) Understanding the origins of these relationships in changing stroke structure gives a new basis for looking at the relationships in Taylor's table more analytically. Many of the relationships here can be found in superficially dissimilar letters in Taylor's table:

A a
B b
C c G g
D d
E e
F f
H h
I i J j
K k
L l
M m
N n
O o
P p
Q q
R r
S s (also the now archaic 'tall s' that looks similar to f)
T t
(Y) y V v U u W w
X x
Z z (and the 3 shape variant)

Thank you for that. I was in a bit of a hurry to make the table and I wasn't really trying to make it a 100% correct or to be completely thorough, just to point out the letters I thought were similar enough on the first glance.

The comparison with the Latin alphabet is a good one. It shows how letters evolve through handwriting and end up completely different in the end. Both Latin and Cyrillic derive from Greek, but at different time periods (around 15 centuries apart), and the evolution of the Greek script thus contributed to the current difference between the two scripts.

By the way, anyone who wants to see original samples of Avestan calligraphy (and a large portion if not themajority of those that are known of) should go to the site of the Avestan Digital Archive organised and hosted by the University of Salamanca:

http://ada.usal.es/

It contains high quality digital images of manuscripts hosted in libraries in numerous countries in Europe and Asia. The manuscripts are written, unsurprisingly, in Avestan script, ranging from clumsy and inept near-scrawls in one or two mss to the most exquisitely executed calligraphy in others. Most also contain annotations and/or commentary in Indo-Persian script, earlier varieties of Nagari similar to Jainanagari, and Gujarati script from older varieties to modern 20th century versions, some of these also containing exquisite calligraphy.

I discovered this source as part of my searches for original data on older varieties of Gujarati and Nagari script, which I needed to flesh out my work on the early Gujarati origins of the Indonesian-Philippine post-Nagari scripts. It's well worth looking through, even if it's just to gain an appreciation of the beauty of some of the calligraphy.

Well for the ones that show up they are amazingly beautiful and is making me wonder what the script would look like today in modern times with all the typefaces.

*Goes on her own search for even more*

I downloaded the book you mentioned and I find his theory interesting. Is there any new research on the subject of Arabian origin of Brahmi?

For Delodephius-

There's no renewed research so far on a South/Old North Arabian connection for Brahmi. As far as I know, I'm the only person who has noticed the otherwise unrecognised connections in the new data uncovered since Taylor's time.

So far, what seems to be the consensus is summarised in Richard Salomon's 1995 review article "On the origin of the early Indian scripts", which is available online at JSTOR (for free if you belong to a subscribing library). This is basically that (1) no recovered texts in Brahmi can be dated with confidence to before the mid 4th century BCE and (2) although no systematic relationship can be established between its letter shapes and those of any other single script, it seems likely it was patterned after a Semitic script, probably (for geographic and historical reasons) Aramaic. Arguments like these aren't generally considered sufficient to establish a relationship between spoken languages, so I am naturally skeptical of the established hypothesis (and also of hypotheses of native invention or development from the Indus Valley symbols).

I think the Safaitic hypothesis is worth a second look because of systematic correlations in structural elements between Brahmi-Old North Arabian letter pairs, precisely describable correspondences in certain letters (similar to the change from V to u etc.), and near direct correspondence between certain 'alif, 'ayn and ghayn variant shapes in Safaitic and Taymanitic-Thamudic and the Brahmi a, i, e, u and o letters. The argument is not overly complex but it does involve a bit of theory, which I am developing in my paper on the Nagari-Gujarati origin of the Sumatra-Sulawesi-Philippine scripts. Once I have finished this paper (which is very long and complex), I will have the basis to deal with the Old North Arabian-Brahmi theory.

By the way, I was wondering about the meaning of your online name. I managed to find the explanation online and sent you a Facebook friend request. (It reminds me of a guy I worked with long ago, a Chinese immigrant from South Vietnam, who called California "Kafitolia" and liked to dance "lat-si-ko" – which I imagine is just the pronunciation of the characters for disikuo (?) in his southern Chinese language.)

For Serali-

Yes, some of the calligraphy is very nice, and the best seems to be in the Meherji Rana library collection in Gujarat (though the Iranian Majles Library also has one or two nice specimens also). I am particularly interested in the Gujarati and Nagari script found in these documents because of my research. Some of the texts have the nicest Gujarati calligraphy I have seen, and they show a lot of variation compared to the shapes that became standard after the 19th century. You can actually see how the modern long ‹i:› letter developed from the older shape, and how the modern short ‹i› letter actually developed from ‹i:›, replacing the original short ‹i›. There is also a very different ‹e› letter related to the Nagari one, which was replaced in modern Gujarati script by ‹a› with the bound ‹-e› vowel mark that appears on consonant letters.

I derive my username from the part of town I live in called Delodefia. It is a corruption of the name Philadelphia. Many Slovaks who went to work in the US brought back with them names of the states and cities they went to and often those names ended up corrupted (old people do this even today with foreign name, my grandfather for example calling my dog Abi or Obi instead of Odi). I remember for example that they called California as Kalafonia, New York as Nev Jork. There is a corruption of other words like encyclopedia > anciklopandia, asfalt > flaster, nectarine > nagdarin, cholesterol > kalastala. These words in proper literary Slovak are identical to their original form.

Here's an interesting example in print:

http://en.wikipedia.org/wiki/File:A_Pag ... _%27Dabistān-i_Mazāhibm%27_prepared_and_printed_by_Fardunji_Marzban_(1815).jpg

Kiwehtin, I would love to hear more about your system of description. I was trying somewhat to formulate such a system myself after seeing someone who was systematically avoiding upper left corners in his handwriting (mostly majuscules of course). I feel even a fairly minimal set of 'subgraphemic components' (such as upper left corners) could be used to test the hypothesis that individual orthographies are characterized by use of a somewhat small set of such components (the same way spoken languages use a small set of phonemes). Of course such components would also be good for systematizing patterns of change within orthographies. Naturally the sort of borrowing which occurs when a new orthography is invented would behave a bit differently than change in one system over time, but obviously there are regularities there.

It's not my field or anything; I'm an undergraduate computer science student. But I'm very interested. When might your work be published? Where are you doing your research?

Hi "dranorter",

Thanks for asking about this. What I'm working on has two levels of description, basically. One level uses a vocabulary of geometric elements (head strokes, stems, bodies, curls, loops, counters etc.) derived in part from typographic terminology. The more basic level describes strokes in terms of defining (target) points that are (a) abrupt, defining a clear endpoint or change of direction (stop points) or (b) indeterminate targets defining a general area where a stroke slowly curves into a different direction (glide points). Defining points are the loci where a range of different processes can take place including gliding (e.g. what happened in the change from v to u) or its opposite, stopping, or fragmentation, where two strokes can separate so they are no longer contiguous.

Setting down the definitions and examples and showing how this all works is rather long and involved, but that is what I am working on in a manuscript I'm writing about the origins of scripts of Sumatra, Sulawesi and the Philippines in early Gujarati script. I imagine I might have the manuscript complete sometime in the summer, but I could send you a draft of the relevant parts if you're interested.

I'm working on this at home, which is where I do my work.