Panel 41, and what a flat composition actually looks like
We matched five base materials for tenacity to see what would happen. What happened was nothing, for six days, and then everything at once.
The idea was not clever. If a fragrance is a set of things leaving at different rates, we wanted to know what was left of it when the rates were the same. So we took five base materials that our own readings put within a few hours of each other — oakmoss, cedar, benzoin, vetiver, ambrette — trimmed the outliers with dilution until all five read at 140 hours, and dipped a rack of eight strips.
What the panel wrote down
Nothing, mostly. At thirty minutes, six people wrote a version of dark, resinous, one thing. At eight hours, the same. At twenty-four hours, the same, and by then two of the six had stopped bothering to write a sentence and were putting a tick.
That is the finding, and it is duller than we wanted it to be. A composition where nothing leaves before anything else does not smell like a chord. It smells like a single material you have not met before, and the reason is that the only information a nose gets out of a pyramid is the order of the departures. Remove the order and you remove the fragrance; what is left is an ingredient.
And then it went
Between the fifth and sixth reading it emptied. Not thinned — emptied. Four of six could name it on the Saturday and none of six could name it on the Sunday, and the strips read of cellulose and the room.
If you look at the chart on the blotter page you can see why in about two seconds, which is the argument for drawing these things at all. Five lines that start together and stop together is not a shape. Put the same five materials at their honest tenacities and you get a staircase, and the staircase is the product.
What we kept
Panel 41 was abandoned after a fortnight. Two things survived it.
The first is that ambrette at that dilution is doing more than we had credited it with; it is now in the store at 1% as well as 10%, which is a nuisance and worth it.
The second is that we now run a matched-rate strip alongside any new base accord as a control. If the new accord reads no more structured than the flat one, the accord is not an accord yet. It has caught two formulas since, including one we were about to price.