Three Dry Lands and Seven Rivers: The Desert the Rig Veda Never Describes

A Map With Three Blanks

Savitṛ, the god who sets the world in motion each morning, is credited in one verse with taking in everything at a glance. The poet wants to say how much he sees, and he does it with a list:

aṣṭau vy akhyat kakubhaḥ pṛthivyās trī dhanva yojanā sapta sindhūn

“He has looked out over the eight quarters of the earth, over the three dry lands by their stretches, over the seven rivers.”

RV 1.35.8, translation mine, checked against Griffith (1896). [1][2]

Every student of the Rig Veda knows the last item. The seven rivers (sapta sindhavaḥ) are the poets’ home address, and a small library has been written about which seven they were. The phrase just before it gets almost no attention. Trī dhanva: three dry lands. Not three mountains, not three seas, not three kingdoms. Three stretches of country defined by what they lack. The god doing the looking matters too: Savitṛ is the one deity who turns up again in the dry-land verses (RV 2.38.7 has him assigning even the beasts of the dry country their share of water), and his own essay, on the impeller of the Gāyatrī, shows why a god of the moving sun is the natural one to survey a landscape.

A people that lists its deserts beside its rivers, in the same breath and on the same footing, has looked hard at both. So the obvious next step is to go and read what the Rig Veda says about its deserts. There is a word for them, dhanvan. It turns up in the text again and again. And here the trail does something unexpected: the poets never describe a desert at all. There are no dunes, no heat, no mirage, no sand. What they do instead, nineteen times in ten books, turns out to be a pattern with a logic of its own, and the pattern says more about how these people lived than a description would have.

What follows is an attempt to read those nineteen verses as a set, rather than one at a time as footnotes to river geography. The method is simple: find every instance, sort them by what the poet is doing with the dryness, and see what the sorting reveals. Along the way we will meet a word that is also a bow, a god who makes deserts walkable, a fire that is compared to a roadside water-stall, and an absence that is easy to miss and hard to forget.

A Word That Is Also a Bow

First, a caution about the evidence, because the word makes it harder than it looks. Dhanvan means a bow in the Rig Veda, and it does so often. In RV 6.75.2, the hymn to the weapons of war, the instrumental dhanvanā is the bow in the soldier’s hand: “with the bow let us win cows, with the bow the contest.” A book earlier, in RV 5.53.6, the identical form dhanvanā appears in a verse about rain: the downpours “go along the dry land.” Same spelling, same case. One means an archer’s weapon, the other a stretch of parched country.

Dictionaries list both senses under the one headword, and I could not find an agreed answer to whether the bow and the dry land are a single word that drifted or two unrelated words that collided. [3][4] For readers the practical consequence is that a keyword search overcounts badly. Running through the text for every form of dhanv- and discarding the bows, the proper names (the Sudhanvan family of craftsmen who become the Ṛbhus) and the verbs for flowing (dhanv-, “to run”; Soma “runs” through the filter over and over in Book 9), I am left with nineteen verses I would classify as dry land. Two of them, RV 5.7.7 and RV 10.115.6, I include with less confidence than the rest, since their syntax is compressed and the translators disagree.

Methods note. Counts of this kind depend on the text file and the judgment of the person sorting it. I worked from the Sanskrit text and Griffith’s English as hosted on this site, and checked each hit by reading the verse. Another reader might drop a verse or add one. The pattern described below does not hinge on any single case; remove any three and it still stands.

Here is how the nineteen sort out.

Notice what is missing from the table. There is no row for “what the desert looks like”, none for “who lives there”, none for “the god of the desert”. Nobody is born in these verses. Nobody herds, farms, trades or settles in them. The dry land is always somewhere you arrive from, pass over, or watch being rained on.

The Day the Dry Land Was Flooded

The largest group is also the most direct. Six verses are, in one way or another, about a god arriving with water. The hymn that does it most vividly is addressed to Indra, and the verse is wonderfully economical:

dhanvāny ajrāṃ apṛṇak tṛṣāṇām

“He filled the dry lands and the open plains for those who thirst.”

RV 4.19.7, a single pāda, translation mine, checked against Griffith (1896). [2]

This verse belongs to an Indra hymn of Book 4 in which his victory over the serpent Vṛtra is told as the release of imprisoned waters (see the Vṛtra myth and its Indo-European cousins); Macdonell noted long ago that releasing the rain is Indra’s chief business. [12] In the same verse the “young women who know the order”, the rivers, have already been sent flowing. Now the water reaches the one place that never had any, and the poet chooses the verb pṝ, “to fill”. Filling is what you do to a vessel. The dry land is being treated as a container that has been standing empty.

The same instinct drives RV 4.17.2, where at the birth of Indra’s fury heaven shakes, earth shakes, “the mountains were shaken, the waters ran, the dry lands were moistened.” And it drives a verse in the next hymn-group about the Ṛbhus, the craftsman-spirits who sleep for twelve days as guests of the god Agohya (“the one who cannot be hidden”) and repay the hospitality by remaking the world:

sukṣetrākṛṇvann anayanta sindhūn dhanvātiṣṭhann oṣadhīr nimnam āpaḥ

“They made good fields; they led the rivers. On the dry land plants stood up, and the waters went to the low ground.”

RV 4.33.7, translation mine, checked against Griffith (1896). [2]

(The twelve-day stay is a hospitality story at heart; for how the Rig Veda handles guests, see The Guest Who Never Leaves, and for the Ṛbhus themselves the artisan gods.) The order of events inside the verse is the point. Fields first, then rivers, then plants on the dry land, then water seeking its level. It reads like a time-lapse of irrigation, the sequence a farmer on the edge of the dry country would recognise.

But the most revealing line in this group is a verse in a hymn to Parjanya, the rain-god:

avarṣīr varṣam ud u ṣū gṛbhāyākar dhanvāny atyetavā u

“You have rained your rain; now hold it back. You have made the dry lands fit for crossing.”

RV 5.83.10, translation mine, checked against Griffith (1896). [2]

Read that twice. The rain does not merely end the drought. It makes the dry land passable. Atyetavai is an infinitive of purpose built on the same verb the poets use for crossing rivers: the dry land is a barrier that rain dissolves. The first half of the verse is a polite request that the god stop, which tells you the poet had also lived through the other kind of rain, the kind that does not know when to quit. These are people who negotiate with weather. (More on Parjanya’s Indo-European relatives in the Parjanya essay.)

One more thing about these six verses: the god acts, and the desert is the evidence. Nobody is mourning the dryness. The dry land is where the rain is seen to land. It is a stage, not a place.

Aside. A tempting reading is that the dry land is thereby treated as evil or hostile, the way Vṛtra’s drought is. The text does not quite say so. The vocabulary of enmity (druh, ahi, dasyu) never attaches to dhanvan. It is not a monster or a curse. It is a condition that is waiting for a verb.

The Drinking Hut in the Waste

Then there are the similes, and here the poets reach for what they actually knew. When they want an image of urgent relief they choose, three times, a person in the dry country and water coming to him.

The most beautiful is addressed to Agni, in the opening verse of a hymn to him:

dhanvan iva prapā asi tvam agne

“You, Agni, are like a drinking-place in the dry land.”

RV 10.4.1, translation mine, checked against Griffith (1896). [2]

The word prapā matters. It names a stall or hut where water is dispensed to travellers; later Sanskrit uses it for the public water-sheds that kings and pious donors set up along roads. The Rig Veda is saying that its fire god is to a prayer what a roadside water-house is to a man crossing the waste. He is the place where you stop, where somebody looks after you, where you are expected. The comparison also quietly records a piece of social infrastructure: a world in which such stalls existed, and where a hymn could assume every listener had stood in front of one with a dry throat.

Two more similes carry the same weight in the other direction. In RV 6.34.4 the Soma and the hymns that reach Indra are compared to waters coming to a person in the dry country: janaṃ na dhanvann abhi saṃ yad āpaḥ. The syntax is compressed and translators spread it differently, but the picture is clear and Griffith’s “like water brought to men in desert places” is not far off (Geldner’s German, the benchmark Griffith is usually checked against, takes it the same way in outline [15]). And in a Soma hymn of Book 9 the simile turns grim:

dhanvan na tṛṣṇā sam arīta tān abhi soma jahi pavamāna durādhyaḥ

“As thirst comes upon a man in the dry land, so, Soma, as you are cleansed, strike down those of evil thought.”

RV 9.79.3, translation mine, checked against Griffith (1896). [2]

Here thirst itself is the weapon. The poet wants an image of something that overtakes a person and cannot be argued with, and chooses the thirst of the dry land. It is a hard line and a revealing one: this is thirst as a person who had felt it would describe it. Not romantic. Not poetic. Inescapable.

Set side by side, these let you see how the poets reasoned about water. In every case the dry land is the measure of how much a gift is worth. The idea that water given to a thirsty person is worth more than water given to an indifferent one is built into the whole sacrificial economy of reciprocal giving, and a reader of the essay on debt and obligation in the Rig Veda will recognise the logic.

Roads Through Nowhere

The five crossing verses add a different kind of evidence. Here the dry land is not being rained on; it is being traversed, and the way the poets talk about the crossing tells us about travel.

The simplest is a blessing. In a hymn to the All-Gods, the Maruts are asked in RV 10.63.15 for wellbeing across a whole itinerary of danger:

svasti naḥ pathyāsu dhanvasu svasty apsu vṛjane svarvati

“Well-being for us on the paths across the dry lands, well-being in the waters, well-being in the contest for the light.”

RV 10.63.15, translation mine, checked against Griffith (1896). [2]

This reads like a traveller’s insurance clause. The poet works through the kinds of ground a Vedic man might cover: roads that run over dry country, water crossings, and the clash that ends the journey. The dry stretch comes first. Evidently there were enough pathyā dhanvasu, “road-ways in the dry land”, to merit their own line.

Then there is an idiom, which appears three times and is the most charming detail in the whole set. The poet wants to say that someone will get past a difficulty quickly and cleanly, and the picture he reaches for is crossing a dry land:

• RV 3.45.1: “Let no one hold you back, Indra, as fowlers hold a bird; go past them as over a dry land.” (ati dhanveva tān ihi.) The god’s horses in the same verse are “peacock-haired” (mayūraromabhiḥ), a detail the poet could not resist.
• RV 10.93.6: the helped man “strives toward great wealth, through troubles as over dry lands” (ati dhanveva duritā).
• RV 6.62.2: the Aśvins “cross the waters, the dry lands, the plains” (dhanvāny ati yātho ajrān), in a verse about the divine twins traversing every kind of terrain on their way to the sacrifice.

The idiom ati dhanva, “over the dry land”, is a way of saying “swiftly, without stopping”. That tells us something about what the Vedic imagination thought a dry land was: a place you do not linger. Nobody camps there. You get over it as fast as your horses can go. The Sanskrit verb ati-i, “to go beyond”, is the same one used when the poets cross rivers at fords (compare the ford before pilgrimage) and when they speak of passing out of anxiety into the wide space of safety (see the narrow place and the wide). The dry land belongs to the same family of obstacles. Trouble is a desert you have to cross.

The fifth crossing verse is the one I would most like to have a camera for. In a short hymn to Agni whose refrain asks him to carry the singers across their enemies, the god is described like this:

yaḥ parasyāḥ parāvatas tiro dhanvāti rocate

“Who from the farthest distance shines across the dry lands.”

RV 10.187.2, translation mine, checked against Griffith (1896). [2]

A fire seen from the far side of an empty land. Anyone who has watched a camp fire from a mile out on dark open ground, or a lamp in a window across a plain, knows exactly how this works: you cannot make out the person or the house, only the light, and the light means somebody is there. In a world that distinguishes the dry country from the settled river-land, a fire visible across a dry land is the first sign that you are nearly home.

Fire in the Dryness

That makes five of the nineteen verses in which Agni appears, which is more than any other god and an odd number for a god whose most famous element is not dryness at all. Look at them together:

The first and the third are the surprising ones, because they give a god of burning the powers of a water-source. Fire that makes a stream. Fire that is a place to drink. This is not a muddle. The Rig Veda has a name for the figure who contains both: Apām Napāt, “child of the waters”, the fire that lives in water, a puzzle I looked at in the essay on the fire in the water. What the dry-land verses add is a practical, almost agricultural context for that paradox. In a dry place, a fire is the one thing that lets you stay at night; it is what the stall-keeper keeps going. And the other verse, RV 6.12.5, gives the other face: grass fire sweeping across tinder-dry ground, fast, “like a thief” that nobody catches. A dry land is where fire runs. The poets knew both.

graph TD
    A["Dry land: no water"] --> B["Impassable, uninhabited"]
    B --> C{"God acts?"}
    C -->|Rain| D["Water fills it: 4.19.7, 5.83.10"]
    D --> E["Plants stand, route opens"]
    C -->|Fire| F["Light seen across: 10.187.2"]
    F --> G["Shelter, a place to drink"]
    C -->|Neither| H["Cross fast: 3.45.1"]
    H --> I["Dry land as image of trouble"]

This is, in compressed form, how the poets think about dryness: not as a feature of the land but as the state of the world before a god’s act. The diagram has no branch for “stay there”.

What Is Not There

Now the absences, which are what first made me stop and check the corpus twice.

A full-text search of the collection for the common Sanskrit words for sand (sikatā, vālukā) comes up empty. The word maru, which later Sanskrit uses for “desert”, is absent in that sense; the hits for marut are the storm-gods, a different matter altogether. [3] Nor is there a word for a mirage, though the essay on māyā as a form-making power shows how much the poets could do with illusion when they wanted to. No desert people are named. No desert is the home of a particular god. No hero marches across one to a battle and tells you how hot it was.

Camels do appear, almost entirely in the gift lists of Book 8 and once in a simile in Book 1 (the story of the camel as an accounting entry is told elsewhere). They are never put in a desert. The two sets of words never meet.

What do we make of that? There are three readings, and the evidence does not permit me to choose confidently among them.

1. The poets lived on the river side. The compositional heartland is the Punjab and its neighbours (the standard reconstruction is in Witzel’s survey [14] and the Vedic Index [13]), where the problem is too much water or too little, not wasteland. The dry lands were a boundary condition, known by report and by the occasional trip, rather than a place anyone had a vocabulary for.
2. The genre forbids it. Hymns to gods are not travel writing. A text whose job is to praise Indra for releasing water is not going to waste lines on describing the place the water is released into. The desert is background to the miracle.
3. They had no need. For a pastoral society the useful categories are water and grass, and “dry land” is simply the absence of the first. You do not need a rich vocabulary for something you are trying to avoid.

None of these is exclusive, and the third matches what later texts do. When Sanskrit literature does start naming the desert as a kind of terrain, it does so in the language of military engineering. The Arthaśāstra, in its chapter on building forts, lists four natural types of stronghold: those protected by water, by mountains, by desert (dhānvana) and by forest. Manu’s list of fort types, at 7.70, opens with the dhanvadurga, the desert fort. [5][6] The word is the Rig Veda’s dhanvan, and its function is exactly the one the Rig Vedic traveller would have recognised: a stretch of country so dry that crossing it costs the enemy dearly. The desert became a defensive asset once somebody was in a position to build a fortress in it. In the hymns it is still just the place nobody wants to be caught.

Which Dry Lands? What Archaeology Allows

The obvious temptation, after counting three deserts in 1.35.8, is to put them on a map. I would resist it. The verse gives a number, with no direction, no name and no distance (the word yojanā in the line probably means “stretches” or “yoke-lengths” here rather than a measured mile; for that unit see the essay on Vedic distance). We can say that the poet knew of more than one dry region and thought of them as a set. We cannot say which.

What archaeology and environmental science can supply is the physical background. The most likely candidate for a dry land adjoining the river country is the area now called Cholistan and the Thar, east and south-east of the Indus, where the beds of the ancient Ghaggar-Hakra system run out. Survey work along the dry Hakra bed has recorded hundreds of Bronze Age sites, most of them abandoned by the time the Rig Veda was composed. [7][8][9] Why they emptied has been a long-running argument. A 2012 study of the sediments by Liviu Giosan and colleagues argues that the Ghaggar-Hakra rivers were monsoon-fed, not glacier-fed, and that as the summer rains weakened after the third millennium BCE the rivers shrank to seasonal channels. [10] A separate line of work on lake sediments in north-west India has found evidence of an abrupt weakening of the summer monsoon around 4,100 years ago. [11]

If that story holds, then by the time the Rig Vedic poets were composing, the dry lands next door were not an eternal feature but the visible aftermath of a lost water system. That would give the line in RV 4.19.7, the filling of dry lands “for those who thirst”, an edge we should not push too far, since the date of the hymns is itself disputed (the debate is surveyed in how old is the Rig Veda). But it is worth recording that a poet who describes dry land becoming passable after rain was not inventing the problem.

Aside. The question of how the Sarasvatī relates to all this is a separate and heavily contested subject, treated in the satellite and geology essay and in the river atlas. Nothing in the dry-land verses requires or refutes any particular identification. No verse that mentions dhanvan mentions the Sarasvatī.

The Poetry of “Until”

Step back from the nineteen verses and a single habit of mind comes into view. In the Rig Veda, dryness is never the final state of anything. It is the first clause of a sentence whose second clause is a god’s name. The dry land is filled, or crossed, or lit up from a distance, or turned into a place with a stall and a drink. It is an until: dry until Indra, dry until Parjanya, dry until the fire is seen.

This is easy to read as a religious reflex, and it is one. But it is also a quiet piece of ecology. People whose entire livelihood depends on rain arriving on time will grammaticalise dependence. The poets did not need a doctrine of divine providence to explain why a dry land might become green; they had watched it happen, and the verb they found for it (pṝ, to fill; sṛj, to release; ati-i, to pass over) is the verb of a person describing a thing that has just been done to the landscape, with the dust still settling.

For a modern reader there is something else here, and it has nothing to do with theology. We now talk about climate as a thing that happens to us, drought and flood as shocks to be managed. The hymns do something different. They treat water as an event, a gift that arrives or fails to, and they respond to the failure not with description but with address: speak to whoever can make the dry land passable. It is a very old way of living with uncertainty, and the thing it produces is not fatalism. It is a habit of asking.

If you want to read the set yourself, put RV 4.19 and RV 5.83 side by side. In the first the god is a warrior who breaks open a world and lets the water out; in the second a poet is half-flattering, half-nagging a rain-god into stopping. They are the same event told from two sides, the thunder and the farmer. Then turn to RV 10.4 and notice the roadside stall in the middle of a hymn about the god you light each morning. Few places in world literature are so quietly practical about holiness. For what water meant when it had to be lifted by hand, the essay on the Rig Veda’s wells is the natural companion, and the cartographers’ essay shows how the other half of the 1.35.8 list, the seven rivers, has been argued over.