The leopard'swindow.

Sunil parked the jeep at four-forty and switched the engine off. We asked, quietly, what we were waiting for. He said only, an hour. The rain began at five. It was the heavy sort of rain that lifts the dust into a smell of iron and wet stone and makes the birds start a second time.

By five-forty the rain stopped and the light in the dry brush turned gold. Sunil had not moved. He pointed once, with his chin, to a low rock at the edge of the water-hole. There was nothing on it. We waited.

At six-twelve a leopard walked out of the long grass, drank for a minute, looked once at us, and was gone. We had seen, perhaps, ninety seconds of the animal. Sunil had given us an hour of waiting and a small chance of nothing, and the trade was the most generous one of the week.

The cost of patience is the only cost worth paying in this park. We do not pretend otherwise to our guests.

What Yala actually is

Yala sits in the dry south-east, a landscape of thorn scrub, granite outcrops and shallow water-holes that looks nothing like the rainforest people expect of a tropical island. It is divided into blocks, and almost everyone who says Yala means Block One, the southernmost section, which carries one of the densest leopard populations recorded anywhere and, as a direct consequence, most of the vehicles. The other blocks are quieter, less predictable and less visited, which suits some guests considerably better.

The leopards here behave differently from leopards elsewhere, which is the reason the park is worth the trouble. With no lions or tigers above them in the hierarchy, they are the apex animal, and they move in daylight with a confidence their relatives in Africa or India generally do not. An animal that walks out of the grass at six in the evening, drinks, looks at a jeep and leaves unhurried is not behaving unusually here. That is the ordinary version.

The odds, honestly

A sighting is likely and it is not promised. Across two or three drives most guests see a leopard; across a single afternoon, a good number do not, and any operator quoting a percentage is inventing it. What genuinely moves the odds is time in the park rather than luck: two drives beat one by a wide margin, and a full day beats two half days, because it removes the pressure to produce something before the gate closes.

The second thing that moves the odds is the tracker, and the difference between a good one and an average one is not marginal. A good tracker reads alarm calls, the direction spotted deer are facing, the age of a pug mark in the sand, and the behaviour of langurs in the canopy, and from that decides where to be and when. The jeep is only the means of getting to the place he has already worked out.

Why the jeep does not move

Waiting is the method, not an absence of one, and it is the hardest part for guests to accept in the first hour. The instinct on a game drive is to cover ground, on the reasonable theory that more park equals more animals. The opposite is generally true here. A tracker who has established that a leopard is in a particular thicket and will need water before dark is not being idle when he switches the engine off. He is spending the only resource that actually buys a sighting, which is patience, and he is spending it deliberately.

The crowd, and how to get around it

Yala's popularity is a real problem and it deserves to be described rather than glossed over. Block One at peak season carries a great many jeeps, and radio traffic means a sighting can gather a queue of vehicles within minutes of being called in. At its worst this is genuinely unpleasant, both for the guests and for an animal trying to reach water through a line of engines. Anyone who tells you otherwise has an interest in not mentioning it.

There are ways around it and they are all versions of the same idea. Be at the gate for opening rather than arriving mid-morning. Take the drives at the ends of the day and give up the middle. Use a tracker willing to sit away from the radio traffic and work his own animal instead of joining somebody else's. Consider the quieter blocks, or Wilpattu, if a guaranteed volume of sightings matters less to you than the character of them. The version of this park worth having is the early, patient, slightly antisocial one.

When to go

February to July is the window, and the reason is water. As the dry season advances the scattered water-holes shrink and the animals concentrate around what is left, which makes them findable in a way they simply are not when the whole park is wet. The park also normally closes for several weeks around September and October, when conditions are at their most severe, so a journey planned across that gap needs checking rather than assuming.

Within the day, the hours matter as much as the month. The first hour after the gate opens and the last hour before it closes are worth more than everything between them, which is when the light is usable, the heat is bearable and the animals are moving. The middle of the day is for the camp rather than the park, and an itinerary that ignores that is buying jeep hours instead of sightings.

Where you sleep changes the drives more than the vehicle does. The tented camps on the buffer land outside the park put you within a few minutes of the gate, which is what makes a genuine dawn start possible rather than theoretical. Hotels further out add an hour of road at both ends of the day and quietly remove the best light from the itinerary. The camps are also the better experience in their own right, since the buffer holds most of the same animals and the sounds outside a canvas wall at night are not something a building reproduces.

Yala, or somewhere else

Yala has the density and Wilpattu has the space, and which of those a guest wants is worth establishing before anything is booked. In Yala you are more likely to see a leopard, and more likely to share it. Wilpattu is far larger and much quieter, its villus give it a different and arguably finer landscape, sightings take longer to come, and when one arrives it is often yours alone. It is also the better park for sloth bear, particularly when the palu fruits in the middle of the year.

If the animal you actually want is an elephant, neither is the right answer and Udawalawe is. Being clear about which animal is driving the trip saves a great deal of disappointment, because the parks are not interchangeable and a week that tries to treat them as one delivers a diluted version of all three. We would rather send a guest to one park twice than to three parks once.

What the hour in the rain bought was not the leopard. It was the confidence that the person driving knew what he was doing and was willing to risk an empty evening to do it properly. That is the thing worth paying for in this park, and it is the only part of a safari that cannot be arranged after arrival.