Basically all recent research suggesting that ancient languages didn't have a word for blue comes from people who don't actually know the languages in question.

The current deluge of misinformation on the subject is the result of two factors: poor communication (leading to misunderstanding), and the prestige of Isaac Newton.

Newton's contribution is the easiest to sort out. The division of visible light into seven colours, in his Opticks (1704), was an authoritative canon for several centuries. (In some quarters it still is, for example if you still think of the colours of the rainbow as ROY G. BIV.) As a result, when someone like Gladstone came to write about colour terms in ancient languages, he took it for granted that Newton's seven colour terms were objective, and that the dividing lines between them were objectively real. Refractive indices are quantifiable, and that has created the impression that everything about Newton's materialist approach to colour is equally objective. But that just doesn't work with linguistic categories.

In other words, Newton's prestige is responsible for the perception that 'blue' is an objective category of colour. The idea of a language 'not having a word for blue' assumes in advance that 'blue' is an objectively demarcated region of the colour spectrum. Which, of course, it isn't.

This is where the communication problems come in. It's very difficult to talk about colour terms in different languages without using colour terms. And if a modern scientist uses English colour terms to talk about another language, that prejudices everything in advance.

So here's a remedy. Instead of thinking of colour terms in terms of how they translate into other languages, I invite you to think of colour terms in terms of how they map onto something like the Munsell colour system, or the colour palette in your favourite image editing programme. (This mapping doesn't exhaust the nature of colour in language -- more about that at the end -- but it's a useful stepping stone.)

If you do this, you'll see that the English term 'blue' occupies a particular bounded region of the colour palette. That region is defined by a certain quantitative range of hue, saturation, and lightness values. The exact boundary of the region will vary to some extent between English speakers. But one thing I'll highlight is that the region is enormous. It's much, much bigger than for any other English colour term. 'Yellow' occupies less than a fifteenth of a Munsell hue wheel; 'blue' occupies a whole third of the wheel.

(This will be different if you use a different colour scheme, like the CIELAB colour scheme, but note also that the CIELAB scheme is based on perception and language use by English-speakers. That is, just like Gladstone, it assumes English colour categories in advance.)

Now, let's take an example from my own area: ancient Greek (since I don't know Hindi or Prakrit or Sanskrit). Ancient Greek has three colour terms that map onto the parts of the same region that English 'blue' occupies: kyaneos, glaukos, and lampros. But they're considerably smaller, and they extend past the edges of the region occupied by English 'blue'. Kyaneos occupies a relatively narrow slice of the hue range, but extends into lower-value colours (that is to say darker colours), all the way to black. (What English calls 'black' hair, ancient Greek calls kyaneos.) Glaukos occupies a wider portion of the hue wheel, extending into hues that English would call 'green', but it also has relatively low saturation: it also overlaps with English 'grey'. Lampros roughly to the hue range that English calls 'azure', but it also has much higher value and high saturation: it can be used for sky colour too, but it's more typically used to describe metals that aren't copper, nickel, or gold -- that is, metallic silver, with a perception of belonging to the opposite side of the hue wheel away from orange-red.

It's extremely misleading to think of any of these three colours in terms of English equivalents. If you take it that kyaneos = 'dark blue', glaukos = 'grey', and lampros = 'shining' (which are standards that translators often adopt), then you're using the English categorisations. Using the English terms to talk about the actual scope of these terms makes communication difficult and tends to put communication askew right from the start.

We work out what the terms mean, by the way, mostly by looking at what objects the colour terms are applied to. In the case of glaukos, it can refer to vine leaves, an eye colour, and the sky; kyaneos gets used of hair, dolphin skin, rainclouds, and seawater. There are a few passages in ancient sources that talk about colour terms and pigments, which also helps a bit.

Now, with this in mind, you can just as easily turn things around. While it's not strictly false to say 'ancient Greek has no word for blue, therefore the Greeks had no word for sky colour', it's extremely misleading: you can just as easily say 'English has no word for glaukos or lampros, therefore English has no word for sky colour'.

It's more true to say: 'ancient Greek has multiple colour terms that overlap with the region of the colour palette that English calls 'blue'. I can't answer for ancient Indian languages because I don't know those languages, but my default stance would be that something comparable applies there.

Now, I mentioned that Munsell dimensions don't exhaust the linguistic nature of colour. There are several other factors that can define colour terms, but they're not as easily quantifiable: things like metaphor (what does 'green fear' in Homer mean? answer: it doesn't mean anything, because 'green' is a modern English word); saliency (which colours are humans hardwire to pay attention to?); colour event (the subjective nature of the perception of a colour in a particular context); and glitter effect or specularity. I mentioned that Greek lampros is typically metallic: it seems that the specular nature of metal surfaces is important to it. The same goes for Greek porphyreos, which is often translated as 'purple' and gets treated as a problem, but is much less of a problem if you plot it on a Munsell palette (it occupies a huge range of the hue wheel, as English 'blue' does) and recognise that it incorporates a shimmer effect as well (it gets used of pigeon neck-feathers).

Any or all of these factors may play into the nature of colour in ancient India. To properly look at the colour term system there, you'd need a corpus-based approach that looks at the use of colour terms in their context, and in their original language; NOT a dictionary-based approach that translates colour terms into a modern language. Colour terms are badly studied precisely because they're so easy to miscommunicate. A scientific study that starts by looking for equivalents for English colour terms, or which uses the CIELAB colour system, is off to a bad start.

Note: I wrote more about the problem in the ancient Greek context in this piece from last year. It gives a bit more bibliography, including references for prominent figures like Newton, Gladstone, and Goethe (who I didn't talk about here). But I'll warn you that I don't know of any really good treatment of the colour term system in any ancient language. Every treatment that I know of falls into the trap of treating them as having prototypical English equivalents.

Just to clarify, Krishna actually means quite literally "black" or "black one." I asked one of my Sanskrit professors once why Krishna is then always depicted (in everything I have seen) as blue, and he didn't have a hypothesis as to why. The Sanskrit word for blue is nila, as in Nilakanta-- Shiva. This is derived form the myth of Shiva swallowing poison, and rather than killing him, turned his throat blue (hence the name nila-blue, kanta- throat).