Banshtala

The boat left the eastern embankment at five-thirty and reached Banshtala by seven.

There were three of them: Arjun, a younger officer from the regional coordination team named Rafi, and a driver who said nothing. Arjun read the case documentation on his tablet, which told him what the community contained in aggregate — population, livelihood breakdown, canopy cover, sensor station coverage — without telling him what it looked like.

It looked like this: a cluster of raised structures along the embankment edge, the walls pale with lime wash, the rooflines low against the treeline behind them. A jetty with two fishing vessels moored and a third being bailed. Cooking fires already out, the morning past that hour. Children at the water's edge, not playing, watching the boat come in with the particular attention of children who don't see official vessels often.

Rafi handled the mooring. He had been here before, twice — he knew the headman, he said, a man named Kamal who would show them the sensor stations. Arjun stepped onto the jetty and looked at the village. He had reviewed fourteen comparable communities in his two years on the western Sundarbans caseload. They were comparable in the data. They were not comparable in the way a place felt when you stood in it.

Kamal was sixty and broad across the shoulders, a man who had spent years on the water and looked it. He greeted Rafi by touching his arm — old acquaintance, not warmth exactly, something more functional. He looked at Arjun the way people look at officials: assessing without expression, waiting to understand what kind of problem had arrived.

They walked the embankment. Rafi pointed out the sensor stations — three of them, the Authority's standard installation, grey housings on steel posts above the tideline. Kamal walked alongside without commenting on them. He had his own reading of the embankment, which he performed in a different register: pausing at intervals to look at the water line, the silting patterns, the roots of the keora at the embankment base. Arjun watched this and said nothing. He did not have a field for it.

The buffer zone's outer margin was a ten-minute walk from the village edge, through a path that the community had kept clear for the harvest season. The canopy closed above them almost immediately — dense in places, thinner where the salt had come in furthest, the species composition shifting as they walked. Rafi described it from the case documentation. Kamal did not confirm or deny.

At the outer margin the trees opened slightly and the light came through. Rafi checked the coordinates against the documentation and made notes. Arjun looked at the canopy.

The woman was at the edge of the tree line, perhaps thirty metres away.

She was working alone. She had a smoker — the kind used to calm the Apis dorsata colonies before the harvest, a metal canister with a bellows that Arjun had seen in documentation photographs but never in use. She was not using it. She was maintaining it: checking the bellows mechanism, the fuel, the seal around the canister mouth. Her hands moved through the check with an economy that came from doing the same task across many seasons, the movements compressed to their minimum, nothing wasted.

While her hands worked, she looked up.

She was reading the canopy above her in the same motion — not sequentially, not first the equipment and then the trees, but simultaneously, her attention distributed between the mechanism in her hands and whatever the canopy was telling her. Her eyes moved through the branches instinctively: the mid-canopy, the upper branches, a section to the east where the light was different. Pausing. Moving on. The way someone reads something they have been reading for years.

Arjun stood for perhaps thirty seconds. Rafi was speaking behind him about the sensor coverage — what the northern station recorded, what the eastern station missed, the gap in the data between them. Arjun said yes at an appropriate moment. He did not move toward the woman. She had not looked at them. The harvest was two months away. She was maintaining the equipment in the meantime because the equipment required maintenance and because she was, he understood, reading the colony's condition through the trees while she did it. Both things at once. Neither subordinate to the other.

He did not write this down. He had no field for it.

Kamal had stopped walking. He was watching the woman the way you watch someone doing skilled work — without intrusion, with a quiet attention that was its own form of respect. After a moment he said something Arjun didn't catch, low, half to himself.

Rafi translated: She's checking whether the season will be early.

They walked back to the boat. Rafi talked about the relocation feasibility parameters on the return path. The community's resettlement eligibility, the compensation structure, the transition support programme. Arjun listened. He looked back once at the tree line, but the woman was no longer visible, or he could not find her from that angle, which was not the same thing.

The boat back to the eastern embankment took forty minutes. Arjun reviewed the documentation on the return. The relocation feasibility table. The ecosystem resilience projections. The confidence interval the model had not yet settled — it would be another six weeks before the case reached its final parameters and landed in his queue.

He looked at the buffer zone coordinates on the map. The community boundary marker. The canopy cover percentage expressed as a number.

He filed the visit report at 4pm from the regional office. Site visit completed. Checklist items confirmed. No anomalies noted.

He did not mention the woman. He had no field for her. He had no language for what he had seen that the report's structure would accept, and so he wrote what the structure required and submitted it, and the system logged it, and the case moved forward in its ordinary way.