If India were still an island, it would be a mid-ocean continent with no Himalaya, no Tibetan Plateau and no Indo-Gangetic Plain, and it would get a much weaker, more equatorial monsoon. That is not far-fetched: India raced north at 18 to 20 cm a year before hitting Asia about 50 million years ago, and if it had moved at its present 5 cm a year it would still be roughly 1,000 km offshore today, by my estimate. The country that exists, with its rivers, rains and 43% of its people in the Ganga basin, is a product of that collision.
- 18 to 20 cm a year: India's speed before the collision, about four times today's 5 cm a year (Kumar et al., Nature 2007).
- About 100 km: the thickness of India's lithosphere, against 180 to 300 km under its old Gondwana neighbours.
- Up to 3,600 km: how far India and Asia have converged since about 52 million years ago.
- Largely absent: the South Asian monsoon in a climate model with all elevated topography removed (Boos and Kuang, Nature 2010).
- 26% and 43%: the Ganga basin's share of India's land and population, all of it resting on Himalayan silt.
Most "what if" questions about geography are idle. This one is not, because the geologists have measured almost every number we need: how fast India moved, why it moved so fast, how the mountains it built stay up, and what climate models do when you delete them. So here is the island India, worked out.
Was India really an island?
Yes, for tens of millions of years. India broke away from the supercontinent Gondwana, which it had been part of until about 140 million years ago, and drifted north across the Tethys Ocean as an island continent until it met Asia around 50 million years ago. The best evidence for life on that drifting island is a fossil found in Kashmir.
The Indian geologist A. Ranga Rao collected truckloads of fossil-bearing rock at Kalakot in Kashmir. Many blocks were still unopened when he died, and in 2005 his widow passed some of them to the American palaeontologist Hans Thewissen. In Thewissen's lab, a fossil preparer chipping at the skull of a small, deer-like animal called Indohyus accidentally knocked off a bowl-shaped ear bone, the involucrum. Only whales hear with that structure. The accident showed that whales' closest relatives lived about 50 million years ago on the shores of India, as the island was arriving at Asia.
The Wire Science: How India and Pakistan fossils rewrote whale evolution, October 2020
Why did India move so fast?
Because its plate is thin. A team from GFZ Potsdam and India's National Geophysical Research Institute measured it with seismic waves.
Using S-wave receiver functions, a seismic method for locating the base of the lithosphere, the team found that India's lithosphere reaches only about 100 km deep, against 180 to 300 km under South Africa, Australia and Antarctica, its former Gondwana neighbours. Palaeomagnetic reconstructions say India reached 18 to 20 cm a year in the late Cretaceous, then slowed to about 5 cm a year after hitting Asia around 50 million years ago. The authors propose that the plume that broke up Gondwana melted off India's lower lithosphere, letting it move faster. The limit: I read the abstract and the GFZ release, not the full paper, and the melting mechanism is a hypothesis.
Read the summary (free)See the data as a table
| Lithosphere depth (km) | |
|---|---|
| India | 100 |
| Other Gondwana fragments, thinnest | 180 |
| Other Gondwana fragments, thickest | 300 |
Source: Kumar et al., Nature 2007 (receiver functions) · South Africa, Australia and Antarctica fall in the 180 to 300 km range.
How long would India take to reach Asia at a normal speed?
Long enough to still be at sea today, on reasonable numbers. This is the core calculation of the post.
The worked derivation: the drift clock
Assumptions, all mine and stated so you can change them: India set off northward about \( t_0 = 100 \) million years ago (when it separated from Madagascar); the ocean it had to cross was \( D = 6{,}000 \) km wide; in reality it closed that gap by 50 million years ago, which needs an average of 12 cm a year, consistent with the 18 to 20 cm peak; the counterfactual India moves at its present speed, \( v = 5 \) cm a year.
- Time to close the gap.
\[ t = \frac{D}{v} = \frac{6000\ \text{km}}{5\ \text{cm/yr}} = \frac{6\times10^{8}\ \text{cm}}{5\ \text{cm/yr}} = 1.2\times10^{8}\ \text{yr} = 120\ \text{Myr} \]
- Collision date. Starting 100 million years ago, that is 20 million years in the future:
\[ t_{collision} = t - t_0 = 120 - 100 = +20\ \text{Myr from now} \]
- Gap remaining today.
\[ D_{left} = D - v\,t_0 = 6000 - (5\times10^{-5}\ \text{km/yr})(10^{8}\ \text{yr}) = 6000 - 5000 = 1000\ \text{km} \]
- Sensitivity. The answer swings hard with the inputs, as the table shows. At 6 cm a year and a 6,000 km gap, India would be arriving about now; at 4 cm a year it would still be 2,000 km out.
| Ocean gap D | Speed v | Time to cross | Collision | Gap today |
|---|---|---|---|---|
| 5,000 km | 4 cm/yr | 125 Myr | 25 Myr from now | 1,000 km |
| 6,000 km | 4 cm/yr | 150 Myr | 50 Myr from now | 2,000 km |
| 6,000 km | 5 cm/yr | 120 Myr | 20 Myr from now | 1,000 km |
| 6,000 km | 6 cm/yr | 100 Myr | about now | 0 km |
| 7,000 km | 5 cm/yr | 140 Myr | 40 Myr from now | 2,000 km |
Use this to check what you just read with your own gap and speed.
When would a slower India hit Asia?
What actually holds the Himalaya and Tibet up?
A deep root of light crust floating in denser mantle. The named theory is Airy isostasy: mountains float like icebergs, so a high plateau needs a thick crustal keel beneath it.
- Balance the columns. For a plateau of height \( h \) with crust density \( \rho_c \) over mantle \( \rho_m \), the root depth \( r \) is
\[ r = h\,\frac{\rho_c}{\rho_m - \rho_c} \]
- Plug in Tibet's average height of over 4.5 km, with textbook densities \( \rho_c = 2{,}800 \) and \( \rho_m = 3{,}300\ \text{kg/m}^3 \):
\[ r = 4.5\ \text{km} \times \frac{2800}{500} = 25.2\ \text{km} \]
- Total crust. Add a normal 35 km crust (my assumption): \( 35 + 4.5 + 25.2 \approx 65 \) km. Seismic surveys find Tibet's crust about 70 km thick, so the simple float model gets close.
That root is made of crust squeezed out of the collision. No collision, no thickened crust, no plateau. The scale of the squeeze is the subject of a much-debated paper.
The authors start from a mismatch: since about 52 million years ago India and Asia have converged by up to 3,600 ± 35 km, but the crustal shortening geologists can find in the Himalaya and Asia is up to about 2,350 km less. Using palaeomagnetic data, they propose that the first, roughly 50-million-year-old collision was of a Tibetan-Himalayan microcontinent, followed by subduction of a mostly oceanic Greater India Basin, and that the "hard" collision with thick Indian continent came only around 25 to 20 million years ago. The limit: the two-stage model is contested, and other groups date the main collision differently.
Read the paper (free)
Would an island India still get a monsoon?
A much weaker one, pushed towards the equator. The Himalaya act as a wall that keeps hot, humid air over India apart from cold, dry air to the north. Remove the wall and the two mix.
The authors ran the Community Atmospheric Model (version 3.5) with modern sea surface temperatures and changed only the topography. With all elevated topography removed, the South Asian monsoon was largely absent: rainfall maxima and the low-level westerly winds were strongly reduced and shifted towards the equator, and the Himalayan rainfall maximum disappeared. Removing only the Tibetan Plateau, while keeping the narrow Himalaya and nearby ranges, left the large-scale monsoon almost unchanged. They concluded that the mountains work mainly by insulating warm, moist Indian air from the cold, dry extratropics. The limit: fixed ocean temperatures mean the model ignores how an island India would also change ocean currents.
Read the paper (free)The worked derivation: why mixing kills the monsoon
Monsoon convection is fed by air with high moist static energy, the energy a parcel carries as heat plus water vapour. The theory behind Boos and Kuang's argument is that the monsoon sits over the peak of this energy near the surface.
- Moist static energy at the surface (height term zero):
\[ h = c_p T + L_v q \]with \( c_p = 1004\ \text{J/kg K} \) and \( L_v = 2.5\times10^{6}\ \text{J/kg} \).
- Indian summer air (my illustrative values: 303 K, specific humidity 0.020 kg/kg):
\[ h_{India} = 1004(303) + 2.5\times10^{6}(0.020) = 304.2 + 50.0 = 354\ \text{kJ/kg} \]
- Dry continental air from the north (300 K, 0.006 kg/kg):
\[ h_{north} = 1004(300) + 2.5\times10^{6}(0.006) = 301.2 + 15.0 = 316\ \text{kJ/kg} \]
- Half-and-half mixing gives 335 kJ/kg, a loss of 19 kJ/kg, which is the same energy as cooling the air by about 19 K. Even a quarter of dry air knocks off about 9.5 kJ/kg. Sensitivity: the drop scales directly with the humidity gap, so a moister north would hurt less.
Today the June to September monsoon delivers a long-period average of 87 cm of rain over India (IMD, 1971-2020 average). An island India would not keep all of it. Two things would survive, though:
- The Western Ghats rains, because those hills are a faulted escarpment raised during the Gondwana break-up, not collision mountains, and they still lift ocean air on the west coast.
- Monsoon onset over Kerala, which depends on the sea and the sun's march north, though the rains would struggle to push far inland. The mechanics are in how the Indian monsoon works and why the monsoon hits Kerala first.

What would happen to the Ganga plains?
They would not exist. The Indo-Gangetic Plain, about 7 lakh km², is built of silt carried down by the Indus, Ganga and Brahmaputra into the trough between the rising Himalaya and the Deccan. No Himalaya means no snow-fed rivers, no trough and no silt.
- The Ganga basin covers 8,61,404 km², 26% of India's land, and in 2001 held about 43% of its people (44.83 crore), according to the Ministry of Environment.
- Snowmelt rivers that run in the dry season would be replaced by rain-fed rivers like those of the peninsula.
- The north would be the edge of the old Indian shield facing open sea, not a plain backed by mountains.
| Feature | India as it is | India still an island |
|---|---|---|
| Highest land | Himalaya, 8 km; Tibet over 4.5 km average | Peninsular hills; Western Ghats survive |
| Crust under the north | About 70 km under Tibet | Normal, about 35 km (assumed) |
| Summer monsoon | 87 cm LPA, reaching the northwest | Weaker, closer to the equator (model result) |
| Great rivers | Snow-fed Indus, Ganga, Brahmaputra | Rain-fed rivers only |
| Population heartland | Ganga basin, 43% of people (2001) | No Gangetic plain to settle |
The map of where Indians live is a map of where Himalayan silt landed.The working of this post
So, what if India were still an island?
- Speed: India's 100 km thin plate let it move at 18 to 20 cm a year; at 5 cm a year it would still be about 1,000 km from Asia.
- Mountains: no collision means no 25 km crustal root, no Tibet and no Himalaya.
- Climate: without the wall, the monsoon in a climate model is largely absent and shifted south.
- Land: no Indo-Gangetic Plain, the land under 43% of today's population.
An island India is entirely possible in plate-tectonic terms; it only needed a normal, thick-rooted plate. At 5 cm a year instead of 18 to 20, India would still be roughly 1,000 km south of Asia, colliding about 20 million years from now on my numbers. That India would have no Himalaya, a weak equatorial monsoon instead of today's 87 cm season, and no Gangetic plain, so it would hold far fewer people and look more like a larger, drier Madagascar.
What's new
- Seismic data from 462 stations suggest the Indian plate is tearing apart beneath southern Tibet near 90°E Communications Earth & Environment, Aug 2025
- IMD forecast the 2025 monsoon at 105% of the 87 cm long-period average India Meteorological Department, Apr 2025
What if next?
- What if every Indian language merged into one? Island India's languages would have had no mountain passes to come through.
- What if India had been one hour ahead all along? A different clock for a country that did reach Asia.
- What if nobody in India honked for a day? Back to the plains, and the noise.
Also on this blog: the wettest places in India. More questions live on the Kyun page.
Questions people ask
When did India collide with Asia?
The first contact is usually dated to about 50 to 55 million years ago. Some researchers, including van Hinsbergen and colleagues, argue the hard collision of thick Indian continent came only around 25 to 20 million years ago.
Why did the Indian plate move so fast?
Kumar and colleagues found India's lithosphere is only about 100 km thick, half or less of its Gondwana neighbours, possibly because a mantle plume melted its base. A thinner plate can be pushed and pulled faster.
Is India still moving?
Yes, at about 5 cm a year towards the north-east, still pushing into Asia, which is why the Himalaya keep growing and earthquakes keep happening along them.
Would India get monsoon rain without the Himalaya?
Some, but far less of the large-scale monsoon. In Boos and Kuang's model, removing all elevated topography left the monsoon largely absent and shifted towards the equator, while keeping just the Himalaya preserved it.
How thick is the crust under Tibet?
About 70 km, roughly double normal continental crust, according to seismic surveys. Airy isostasy predicts about 65 km for a plateau 4.5 km high.
Sources
- ScienceDaily (GFZ Potsdam): India, fastest continent because of thinnest lithosphere (2007) (source for 100 km, 18-20 cm/yr and 140 Ma)
- Kumar et al., Nature: The rapid drift of the Indian tectonic plate (2007)
- Boos and Kuang, Nature: Dominant control of the South Asian monsoon by orographic insulation (2010) (source for the no-topography model results)
- van Hinsbergen et al., PNAS: Greater India Basin hypothesis (2012) (source for 3,600 km convergence and the two-stage collision)
- The Wire Science: whale evolution and the Indohyus fossils (2020) (source for the story)
- Wikipedia: Indian Plate (source for separation around 100 Ma and today's 5 cm/yr)
- Li and Mooney, USGS: Crustal structure of China from deep seismic sounding profiles (1998) (source for the 70 km Tibetan crust)
- Wikipedia: Tibetan Plateau (source for the over 4,500 m average elevation)
- Ministry of Environment and Forests: Status Paper on River Ganga (2009) (source for 8,61,404 km², 26% and 43%)
- Wikipedia: Western Ghats (source for their Gondwana break-up origin)
- Wikipedia: Indo-Gangetic Plain (source for the silt-built plain and 7 lakh km²)
- IMD: Long range forecast for the 2025 southwest monsoon (2025) (source for the 87 cm LPA)
- Liu, Shi and Klemperer, Communications Earth & Environment: The Indian Plate subducting below the Tibet Plateau is tearing apart (2025)
