"cannot be accepted without qualification as Aryan. ... The deviations in the inflection are puzzlingly numerous. ... Again the number of Indo-European words and stems identified in the vocabulary is but small. Finally, the syntax remains essentially un- Aryan... Now if these documents dated from the 14th century AD, few would hesitate to declare that they were written in an Indo-European language and explain the discrepancies as due to the familiar phenomena of decay, assimilation of forms, and foreign borrowing. But the texts... are many centuries older than the oldest written memorials of Sanskrit or Greek. Yet their language diverges from the hypothetical original Aryan tongue far more than Greek or Sanskrit differs from the parent speech or from one another. It is a fact impossible to believe that a truly Indo-European language would look so odd in the 14th century before our era."

“By the end of the fourth millennium B.C. the material culture of Abydos, Ur, or Mohenjo-daro would stand comparison with that of Periclean Athens or of any medieval town. . . . Judging by the domestic architecture, the seal-cutting, and the grace of the pottery, the Indus civilization was ahead of the Babylonian at the beginning of the third millennium . But that was a late phase of the Indian culture; it may have enjoyed no less lead in earlier times. Were then the innovations and discoveries that characterize proto-Sumerian civilization not native developments on Babylonian soil, but the results of Indian inspiration? If so, had the Sumerians themselves come from the Indus, or at least from regions in its immediate sphere of influence?”

“These numerals and divine and personal names are the oldest actual specimens of any Aryan speech which we possess. The forms deserve special attention. They are already quite distinctly Satem forms ; in fact, they are very nearly pure Indic. Certainly they are much more nearly akin to Sanskrit than to any of the Iranian dialects that later constituted the western wing of the Indo-Iranian family. Thus among the deities Nasatya is the Sanskrit form as opposed to the Zend Naonhaitya and all the four gods are prominent in the oldest Veda, while in the Iranian Avesta they have been degraded to secondary rank , converted into demons or renamed . The numerals are distinctively Indic not Iranian ; aika is identical with the Sanskrit eka while ' one ’ in Zend is aeva. So the s is preserved in Satta where it becomes h in Iranian and the exact form is found, not indeed in Sanskrit, but in the Prakrits which were supposed to be post-Vedic. Even the personal names look Indic rather than Iranian. Thus Biridaswa has been plausibly compared with the Sanskrit Brhadasva . If this be right the second element, asva, horse, is in contrast to the Iranian form aspa seen in Old Persian and Zend. ... Finally we know that there existed among the Mitanni at this time a class of warriors styled maryanni which has suggested comparison with the Sanskrit marya young men, heroes.”

“The simplest explanation of the presence of a Centum language in Central Asia would be to regard it as the last survivor of an original Asiatic Aryan stock. To identify a wandering of Aryans across Turkestan from Europe in a relatively late historical period is frankly difficult.”