This presentation is partly based on my published paper  dealing with Khojki script as found in some of the earliest manuscripts that I had access to. Prior to discussing the script, I would briefly like to present its history.
Dr Daudpota and Dr Gulam Ali Allana  have identified the Khojki script with the one used on storage jars, which bear short inscriptions in proto-nagari style of the eighth century A.D. The potsherds have been excavated by Dr. F. A. Khan  at Bhanbhore in lower Sind at a level belonging to the Hindu-Buddhist period, before the advent of Islam in Sind. According to Allana, the Khoja Ismaili began to use this script between A.D. 1051 and A.D. 1351. The earliest use of the script, however, seems to have been for business purposes. The Lohanas and the Bhatyas are probably the people who were first to use it and this was continued by the Khojas and the Memans. So far, the earliest written evidence of the script is from a manuscript dated 1737. We also find a complete Khojki alphabet in this manuscript.
In my study of Das Avatar, I have used two manuscripts in particular. One of which I shall call D manuscript (D Ms) belonged to Pirzada Sayyed Muhammed Nurali Shah of Burhanpur (MP) who is the `Sajjada - nashin' of the Pirana Panthis of the Kandesh area. This manuscript was lent to Professor A. A. A. Fyzee, who in turn lent it to the Deccan College in Poona. This manuscript contains 595 folios each measuring approximately 9.2 by 5.6 inches. The text is mainly in Devanagri script, however there are some Ginans in Khojki. One of the dates mentioned in this manuscript is the Sake year 1737, called Yuva which works out to be A.D. 1815.
The other manuscript used, which I shall call the Kx manuscript (Kx Ms) was in the collection of the Ismailia Association of Pakistan in Karachi. It is the oldest manuscript known to me. The earliest date mentioned is 1793 Samvat which works out to be A.D. 1737. It is a large manuscript of yellowish, coarse country paper. The folio numbering is chaotic, because on many of the pages the numbers have been obliterated, while some sections with a different numbering system have been inserted. This manuscript is entirely in Khojki.
In order to present a comparison of the Khojki script, I shall use a Khojki primer published in 1932 as a reference and compare the script as found in the D Ms. and Kx Ms. I have also included the Roman, Devanagri, Gujarati and Sindhi alphabets in the table. The following notes should be used with the table;
1. The same symbol is used for both the short i and the long i. With consonants, the script only uses the sign for the long i, the "dirghai", and never the sign for the short i, the "rasvai". See note 3 below.
2. The Khojki script system uses the same symbol for the long u and the short u; mostly it is the short u, as in ubha, udharea, upara, ugara, ect. With consonants, it never uses the sign for the long u, the "dirghau", but only the sign for the short u, the "rasvau". There is no distinction between the short u and the final o; e.g.the symbol is used for the initial u and the final o in the word utareo.
3. As in other Sanskrit based script system, e is a separate symbol from a. In Gujarati one uses the a, plus a "matra" sign to form the e. In Khojki e is a separate symbol. It looks like the letter p to which the sign `-' has been added. The e can therefore often be misread for pa, pe or pi. See note 10 below. The script uses the sign `-' with other consonants too, to give the short a as in vicara, or the short i sound as in kia, or the short e sound as in pase.
4. Diphthongs are not found in the Khojki script system.
5. The script system uses three different forms of g; the usual Sanskrit g, which is the same as in Gujarati, as well as what can be read as gr in Gujarati; and what can be read as gra, gre and gri.
6. The script uses two symbols for j (though the second looks as if it is a simplified form of the first), as in jare, juga, aja, ect. It uses the symbol for the word jh in words such as majhara, hajhara, ect.
7. The Khojki script system uses the same symbol for the cerebral t as in words like popata, and th as in words like thama baetha.
8. The Khojki script system as found in the Kx Ms. uses the same symbol for the cerebral d (though this is a rarer sound than the other two), and n as in pana and r as in kirori (I think this last is the commonest of the three).
9. The Kx Ms. uses two symbols for the dental dh. It uses one symbol for words, mostly, though not always, of foreign origin, as in mahadhina, dharia, mahamadha, etc. It uses the second symbol apparently for all the words of Sanskrit origin, as in udhare, dharave, dharea, dhiaea, etc.
10. The letter p in the Khojki script is like the letter p in Sanskrit or Gujarati; but in the Khojki script it is likely to be confused with e, e.g. the words pana, ene and ere are all written in the same way.
11. The Kx Ms. uses the same symbol for both f and ph; e.g. in firman and phala.
12. The Khojki script system as found in the Kx Ms. uses two different types of bh rather confusedly (in Gujarati and Sanskrit there is only one symbol for labial bh; as also in Persian and Arabic. In Sindhi there are two sounds bh and bhh. No pattern or reason is discernible for the use of the two different symbols.
13. There is no y in the Khojki script as used in the Kx Ms. There is y in the Khojki primer but it looks very much as if it has been adapted from the i. In cases where the y would be used in the Gujarati language, the language of the Kx Ms. either drops it, or replaces it with a, e ect. For example, pamea for pamya, puna for punya.
14. There are no sibilants s or s in the language of the Kx Ms. Even the Persian words which should be written and pronounced with sh are changed to s; the most common being the Persian word shah. The Kx Ms. instead uses the saha. Professor N. B. Baloch in his " A short history of the Sindhi language, Hyderabad, 1962, 33f records (from the Kitab al Aghani) the story of how the Sindhi poet Abu Ata as- Sindhi (early ninth century) used to pronounce the Arabic word shaytan as saytan. In the Khojki primer there is the sibilant sh, but it obviously seems to be a mixture of the Khojki s and the Sindhi sh.
15. Of all the Sanskrit based languages, this sound seems to be peculiar to Sindhi and Cutchi. In Cutch, the palatal dy is used in such words as raday (Gujarati raja), vadyare (Gujarati vagade), etc. The only other language that I know in which the palatal dy occurs is Kiswahili, e.g. in the word dyambho
16. The palatal nasal ny occurs only twice in the language of the Kx Ms. though it is quite a common sound in both the Sindhi and Cutchi Languages. Again the only other language that I know in which it is found is Kiswahili, as in the word nyama.
17. There is no z sound in the Kx Ms. In most of the Persian and Arabic loan words the z has changed to j; thus pir zadeh has become pir jadeh. The Gujarati script system does have the z symbol, yet in many Persian loan words where the sound z occurs, the Gujaratis use j instead, as in jira from zira, etc. To site an opposite example N. B. Baloch (op. cit) records the story of a Sindhi woman (from the Kitab al Hayawan c.,800 A.H. of al-Jahzi) who pronounced zamal instead of Jamal.
I hope that these few comments and notes on the Khojki script will help in reading manuscripts.
1 Khaki G., "The Dasa Avatara of Pir Shams as linguistic and literary evidence of the early development of Ismailism in Sind"; Sind through the centuries. Oxford University Press, Karachi, 1981, pp 143 - 155.
2 Allana G. A., "Sindhi Suratkhati"; Hyderabad, pp 19 - 23.
3 Khan F. A., "Banbhore: A preliminary report on the recent Archaeological excava- tions at Banbhore "; Karachi, 1963, pp 29.