As regards the development of the case system, we have to proceed from the very origin of the   language at the isolating stage.

12. The most primitive concept in the case system is the subject. It is the basis of the appearance of a verb, and the most primitive clause consisted of the subject and the verb (s + v): Na ra (an animal goes). Then in relation to some other verb even an object was brought in the middle. Therefore, a little longer sentence was subject + object + verb (s+o+v): Na ma Sa (knowledge gives birth to happiness). Later, when the verb was put before the object, it was a passive voice: bu Sa ci (the light is generated through the water). The concept of other cases gradually came into the mind of the Indus people, and very ingeniously they expressed oblique cases by putting a word at the first position and pushing the subject at the second position. As the language advanced, any word-order in the clause could be possible, and the purpose was specific; e.g. object + verb, mA tara (she) gives milk, pi tara (he) gives food, used to express family Relationship, the subject known from the context. Depending on the context, even a verb could begin a sentence; e.g. gA va (goes into the dwelling). Even the subject and object could be omitted: bI ja (in the water it is born). We find a little difficulty in recognising these underivable words of Sanskrit.

    1. With his piercing eyes Panini had seen the various vowels working for different oblique cases: 3.A  4. e 7.i. But it is too early to see them in the extant Indus inscriptions. And the most striking feature of the isolating Indus was that the plural was shown by repeating the monosyllabic word so many times: Na Na Na.

    2. When the language came to the agglutinative stage, neither the manipulation of the word-order nor showing plural by repetition was possible, because many bisyllabic words had come into being in the language.

13. Then specific syllables began to express various ideas. First of all, the plural began to be expressed by -Sa. By this time, the clause Na ra had come closer and had become a phrase ‘a walking animal’. Later it signified ‘a man’. The pl. form of Nara was NaraSa. Because the verbal idea was already digested in the lexeme Nara (a man), a new verb had to be brought: NaraSa gA (men go). The extension of gA to A-gA (went) and later a-gA-ta (he went) will be explained in the chapter on the verbs.

1. In some cases the pl. NaraSa was strengthened as NarASa. On the other hand, the pl. NarASa was reinforced by another -Sa. They all appear in Sankrit with their reduced forms.

14. The next affix at the agglutinative stage was -ma for the direct object and (b) hya for the indirect object. Later (b)hya was reduced to ya and perhaps on account of this the base a- preceding it was lengthened in Sanskrit; e.g. devA-ya from deva-. The final -m in tubhyam and mahyam is a later development, as inferred from Lat. tibi, mihi.

15. The pl. suffix -Sa was also used for showing possession. But it was extended by -ya to be brought under the case system. The ending -Sya is the most enduring and expansive, because it is found in all IE dialects. It was the first to disappear in Apabhramsa after the pl. -as.

16. The pl. -Sa was bent towards -Su to express locative. But then the a- base was raised to e; e.g. deve-Su from deva.

17. The next affix to develop was -ti for the other oblique cases. Added to the pronominal base i- (this), it still survives as the particle iti (thus), also used as postposition to express object in Sanskrit. Its -t element is still surviving in the ablative sg. forms devA-t and ma-t tva-t, etc. When this -ti went out of use, the abl. sg. generally merged with the genitive in Sanskrit.

18. Thus by the end of the agglutinative Indus, the following case-endings were there:

19. The inflexional Indus may be supposed to have started with this legacy. But the language became accented, and the forms began to be reduced in size. Then naraSa became náras, NarAsa  nárAs, NarASaSa, nárAsas, Narama náram, and so on. At this stage, they were analysed in a novel way where the vowel became the cut-point. Thus náras was analysed as nár-as, nárAs as nára-as, náram as nár-am or nára-m. By this analysis, a sg. form nára-s too was postulated, and -s was taken to be the sg. ending.

20. We are now at the inflexional stage of Indus. At this reproductive stage of the language, new forms began to come out. The pl. nárAs was extended to nárA-n-s to express object in pl. When -ns was conceived as the object-pl. ending, it was applied to i- and u- bases. This -n- element is a typically pl. idea.

21. During the emergence of the inflexional Indus, even new consonantal bases began to emerge, changing the face of the various endings. The endings at the inflexional Indus stage were the following:

1. Some specific endings developed from them: dhanin-A gave birth to -nA for muni-nA, sAdhu-nA. This nA was shortened to na when the base deva- was raised to deve.

2. The little pronominal bases a-, i- and u- at the isolating stage formed small clauses with specific verbs. They are the worn out particles Im, sIm, etc. of the RV. Later they were supplemented by new bases like ma- for the first person, tva- for the second person and ta-, amu- etc. for the third person. While the Indus pronominal forms died or became meaningless, the forms of the inflexional stage are very lively in the Vedic language.

Duals

22. A question arises as to why duals were conceived at all. The eternal pairs of the human body, e.g. the hands, feet, eyes, ears, etc. are remarkable, and they may be responsible for it.

1. As the pl. is formed by repeating a syllable so many times at the isolating stage of Indus, the duals too may be seen in the same context. That is to say, the signs repeated twice at the beginning of a clause may be supposed to be the early duals.

2. For the pl. we have already seen the affix -Sa at the agglutinative stage, but for the dual we are not able to imagine any definite affix. In Sanskrit we see that the dual forms end with A I U e and au, of which I U e are said to be uncontractable with the following vowels. The uncontractableness of I U e raises some doubt about their vowel character and compels us to suppose that the final element of these long vowels was some consonant in the prehistory of the language.

23. The laryngeal H of the Indoeuropeanists is said to have the lengthening effect on the preceding vowel. That is to say, munI was originally muni-H, ripU was originally ripu-H. This H may represent one of the two ha-phonemes of the Indus syllabary. This ha continued to be present in Indus or later till the vowel sandhi had begun to operate in the language, and the vowels remained uncontracted due to the presence of the central ha. It would have been at the inflexional stage that the phoneme ha was first reduced to H and was then lost after lengthening the preceding vowel. It is meant to say that the dual affix at the agglutinative stage was -ha. Even where the preceding Ů of the duals remain uncontracted with the following vowels, that may be supposed to end with the prehistoric H.

24. Among the two ha-phonemes of the Indus syllabary, the first ha (of the ha ya va ra group) is just on the border of the vowels. Obviously, it is voiced and has also the nature of a vowel. This alone can be the father of the laryngeal H, because the vowel preceding it has been lengthened. Also, because it is a consonant, the following vowel has remained uncontracted.

1. This ha could play the role of a dual affix. Our EkAkSra-kos'as do not give the meaning ‘two’ for ha, but that may be taken for granted. The dual forms like kavi-ha kratu-ha would have existed in the language for some time. At the time when the sandhi was operating in the language, they would have remained as kavi-H kratu-H and would not have allowed the following vowel to come near. Later when this H too was lost after lengthening the preceding vowel, the memory of its presence would have avoided the sandhi.

2. With the preceding a this H would have produced au on the following pattern: In the sandhi of devas ha, e.g., Panini postulates that the final s of devas becomes r (P. 8,2,66), then u (P. 6,1,114). A tardy pronunciation of this u after the preceding a would have brought forth -au, which was recognised as the dual ending later on.

25. Thus the dual affix at the agglutinative stage of Indus was -ha which, later reduced to H at the inflexional stage, lengthened the preceding vowel while going away but its memory prevented the possible sandhi.

1. These dual forms with -ha were restricted to subject and object.

2. The fricative voiced ha, following a voiced stop, would have acted as an aspirating agent. The Vedic sindhu (river) is said to have developed from syand (to flow) + u, with the interposition of an h. Even in Hindi: ab (now) + hI=abhI (just now), etc.

3. The voiceless ha would have aspirated a voiceless stop. We have noted that there are instances of only Tha as voiceless aspirates in Indus, others being totally absent. They are seen growing in number in Vedic on a limited scale. For example, sak (for sac) + i has interposed an h in the Vedic sákhi. The Av. has the regular haši for the Indus *haci (=s/hats'i/) from s/hac.

26. The dual endings for the oblique cases are –bhyAm and –os in Sanskrit. They are of very late origin in Indus. Thus –bhyAm was borrowed from the pl. –bhyas by attaching the dual –am from the personal pronouns (Avám, AvAm, yuvám  yuvAm).