An upapada of attendant word tatropapadaṃ saptamīstham [[3.1.92]] which does not end with a tense-affix [[3.4.78]] is invariably compounded with that with which it is in construction. The compound thus formed is tat-puruṣa,
Thus कुम्भकारः 'one who wakes pots,' नगरकारः 'one who makes cities.'
Why do we say 'which does not end with a tense-affix'? Observe एधानाहारको व्रजति 'he goes to bring fuel.'
It might be objected, that the question of compounding with a word ending in a tense-affix is irrelevant; since the anuvritti of the words सुप् सुपा is understood here, so that तिङ् will find no scope. To this we reply that we should infer that the words सुप् सुपा should not be read into this and the last aphorism. The following Paribhasha also arises from this sūtra.
गतिकारको पपदानां कृद्धिः सह समासाचनं प्राक् सुबुत्पत्तेः ॥
\It should be stated that Gatis, Karakas, and Upapadas, are compounded with bases that end with krit-affixes, before a case-termination or feminine affix has been added to the latter.\
The result is that Upapadas and Gatis are rot compounded by sūtras 18 and 19 with case-infected nouns, but they are compounded with primary nouns before a case-termination or feminine affix is added to the latter. The same considerations apply to karakas also. Thus अश्वक्रीती 'a female brought in exchange for a horse,' is formed correctly. अश्वेन क्रीयते सा = अश्वक्रीत ; add ङीष 4.1.50 and we have अश्वक्रीती. If on the other hand, the feminine affix ate had beca added to क्रीत previous to its composition with अश्व the form would have been अश्वक्रीता, and we should have had no base ending with short अ and in that case ङीष् could not have been added by 4.1.50.,