The migration is a probability, not a schedule.

Every February a guest telephones and asks which week in August they should book to see the crossing. It is a completely reasonable question and there is no honest answer to it, and the way an operator handles that gap tells you most of what you need to know about them.

Here is what I tell people. The migration is not an event. It is around 1.3 million wildebeest, 200,000 zebra and 400,000 Thomson’s gazelle moving in a rough annual loop between the southern Serengeti and the Mara, driven by rainfall and grass rather than by a date. The Mara river crossings, which are the images everyone has in mind, are a by-product of that movement meeting an obstacle. They happen when the herds are on one bank and want to be on the other, and they can happen four times in a day or not at all for eleven.

What actually triggers a crossing

After nineteen seasons I would say the honest list is short, and only the first two are reliable.

• Grass on the far bank. Everything else is secondary. If the north has had rain and the south has been eaten out, they will cross.
• Pressure from behind. A crossing frequently starts because animals at the back push animals at the front into the water. This is why large massing is a better predictor than any amount of apparent intent.
• A leader animal committing. Genuinely unpredictable. I have watched four thousand animals stand at Serena crossing for six hours and walk away.
• Disturbance. Vehicles parked badly on the far bank will stop a crossing that was about to happen. This is the part we can control, and the reason conservancy vehicle caps matter even for reserve game viewing.

“Four thousand animals will stand at the water for six hours and then walk away. I have seen it more times than I have seen them cross.”

What the calendar is actually telling you

Our own logs since 2017 give a distribution rather than a timetable. The first significant northern crossings have arrived as early as 28 June and as late as 2 August. The median is around 16 July. In 2024 — the year everyone remembers — the long rains overran, the southern grass held, and the first crossings at the Mara river did not happen until the third week of August, eighteen days behind the median.

That is what a migration calendar should look like: a spread with a median, not a row of months with animals drawn on them. When you see a calendar that says July: crossings begin, you are reading marketing.

How to plan around it

Three things materially improve your odds, and none of them is picking a better week.

Stay longer in one place. Four nights on one conservancy beats two nights each in two. Crossings are a function of being there when it happens, and every camp change costs you a morning and an afternoon.

Sleep on the right side. The crossings run both ways all season. A camp with access to both the Mara triangle and the eastern bank, or a mid-journey relocation built into the itinerary rather than charged as an extra, is worth more than any amount of timing.

Accept the four-hour wait. The single biggest difference between guests who see a crossing and guests who do not is willingness to sit. If you have told your guide you want to be back for lunch, you have told him not to wait, and he will not argue with you at the crossing point.

And if it does not happen

Then you will have spent a week in the highest predator density in Africa, and the crossing was never the best part anyway. The guests who write to us afterwards write about a leopard in a tree at six in the morning, or a Maasai guide identifying a bird by its call, or the week the river rose and everything changed. Nobody has ever written to me about a crossing they saw from among thirty vehicles.

That is not consolation. It is the actual argument for how we build these journeys: put people somewhere excellent, for long enough, with someone who knows it, and the postcard becomes a bonus rather than the point.