The decision

Here's the rule I'm suggesting.

I stop at a time I've chosen in advance. The work will still be there tomorrow, and I'll do it then.

It has seven parts.

Count first. If you haven't done the exercise in chapter 1, do it now, for one week. You need to know your real hours, and you need a look at what the last two hours of the day contain.

Choose the stop. Pick a finishing time for ordinary days. I'd suggest the end of your contracted day plus no more than half an hour. If that feels impossible from where you are, choose a time an hour earlier than you leave now, hold it for a month, and then move it again. The time matters less than the fact that it's fixed and chosen before the day begins.

End the day on purpose. Spend the last ten minutes on a short closing routine. Write down where you've got to, and the first three things you'll do tomorrow. Check tomorrow's diary. Then close everything. This is what makes it possible to switch off, because your mind can let go of something once it's been written down.

After the stop, no work. That means no laptop at the kitchen table and no work email on your phone in the evening. If you can, take work email off your personal phone altogether, or turn its notifications off outside working hours. If a thought about work arrives at nine in the evening, write it on a piece of paper and leave it for the morning.

Keep one full day a week with no work in it at all, and preferably two. A full day without work each week is what the law entitles you to, and on Pencavel's evidence it's worth about a tenth of your output. If the Sunday session goes, replace it with twenty minutes of planning on Friday afternoon.

Make exceptions deliberately. There will be real deadlines, and an occasional long day for a real reason is fine. An exception has three marks. It's for something specific, you decided on it in advance, and it has an end date. A board paper that's due on Friday fits. "It's a busy time" doesn't, because it's always a busy time. If you find you're making an exception most weeks, then the long hours have come back.

Tell people. Tell your manager what you're doing and why, in terms of the work. You might say that you've noticed your late hours aren't productive, that you're going to finish at half past five and protect your mornings, and that you'd like her to judge it by results. Most managers will accept this, and some will be relieved. Tell your team as well, and tell them you don't expect replies to anything outside working hours. If you must write an email in the evening, schedule it to send at nine the next day.

Plans of the "when this happens, then I'll do that" kind make a decision much more likely to survive a pressured afternoon. Peter Gollwitzer and Paschal Sheeran pooled 94 tests of such plans and found that they made people a good deal more likely to follow through. Here are three, which you can adapt.

When it's ten minutes before my stop, then I'll write tomorrow's first three tasks, whatever I'm in the middle of.

When I think "I'll just finish this first", then I'll write down where I've got to and finish it in the morning.

When I reach for the laptop after dinner, then I'll write the thought on paper and go and find Ellie.

Sooner or later you'll slip. You'll stay until eight one Wednesday without having planned to. When you notice, stop, go home, and keep to your time the next day. One late night doesn't bring the old hours back.

Finally, put something in the space. An evening with nothing in it will fill up with work, or with the sofa and the phone. Decide what the time is for. The gala on Thursdays is an obvious start. After that, look at the list from chapter 13 of things you used to do, and pick one.