The past had a smell, and I have to get it right

I write novels set in the past, and I am told, kindly and often, that I care too much about the wrong details. A reader wants to know who married whom and whether the letter arrived in time. I want to know what the hall smelled like when they opened the door.

I have decided I am right and everyone else is being polite.

The thing the paintings leave out

We inherit the past mostly through pictures, and pictures are silent and odourless, which is a mercy for the painter and a trap for the novelist. The great hall in the portrait is clean, bright and still. The great hall on the day it was lived in was none of those things. It was smoke from a fire that did not draw well, tallow, wet wool drying by that fire, animals closer to the house than we would allow, and under all of it the particular staleness of a building that could not be aired the way we air a room now.

If my clerk walks into that hall and notices only the tapestries, I have written a museum, not a morning. What would have reached him first, before he saw a single thing, is the air. Get the air wrong and the whole scene floats an inch above the floor, no matter how correct the buttons are.

Where the research actually lives

People assume the hard research is the dates. The dates are the easy part. They are written down and someone has usually checked them. The hard research is the ordinary sensory life that nobody thought worth recording, because everybody already knew it.

So I read the things that were never meant to be literature. Household accounts. Complaints to the parish. Instructions for getting a stain out of linen. A recipe is a wonderful thing, not for the food but for the assumptions folded into it: what was to hand, what was dear, what you did when the good ingredient was not to be had. A page telling a servant how to sweeten a room tells me, without meaning to, exactly how bad the room usually was.

The dates are the easy part. The hard research is the life nobody wrote down because everybody already knew it.

I am not doing this to show off the reading, though I will admit the notes run longer than the manuscript and I regret none of them. I am doing it because a reader can feel the difference even when they cannot name it. When a scene has a temperature and a smell and a sound the character has stopped noticing, the reader stops noticing they are reading. That is the whole trick, and it hides in the parts of the past that never made it into the portrait.

So yes, I will spend a week finding out what the corridor smelled like, and the marriage plot will wait. The marriage is what the reader came for. The corridor is what makes them believe it happened.

The smell of a city is not optional in Two Minutes to Noon, which follows two women across Tokyo as it burns on 1 September 1923.