To err once is unfortunate, but twice is careless. Which is the charge hanging over the US Department of Agriculture after a second batch of controversial data on corn inventories.
Investors were unsettled when the USDA in June came up with a figure 290m bushels (7.4m tonnes) below their own estimates.
But many have been left incensed now that the mighty agricultural force issued a September estimate 300m bushels above what they expected, forcing crop prices into a sharp correction. Chicago corn prices have lost nearly 8% in two days.
To be sure, traders will always grouse over statistics which turn the market against their dealing positions. But the complaints this time look like more than sour grapes.
The September figure throws up two inconsistencies.
The first requires drilling down. Strip out use of the grain in making ethanol and in selling abroad, on which statistics offer a decent picture, it implies use in the livestock feeding and residual segment of only 523m bushels, according to Macquarie calculations.
That's a weak number. It's down 25% year on year, to the lowest since at least the mid-1980s, the bank says.
During a recession, a decline of that scale might have appeared feasible. But it looks unlikely when livestock farmers are enjoying better times and many are scaling up. The number of cattle on American feedlots was 3% higher last month than in September last year.
If that's something of a smoking gun, the second challenge to the stocks figure catches the USDA red handed - simply that it was so far from the last guess.
If analysts appear wayward in being 300m bushels out on their estimate, the department's last estimate for September 1, issed on September 10, was out by a further 20m bushels.
No-one is saying the USDA's job is easy. But the size of the gap between two numbers issued just three weeks apart, is of more than academic concern.
A glance at the implications on futures markets shows why.
In the intervening fortnight between the two estimates, Chicago's front month contract alone traded well over 2m futures. With each contract covering 5,000 bushels, at roughly $5 a bushel, well, that's a stack of value to be traded on imprecise data.
And that excludes the other corn lots, and contracts in other crops such as wheat, which have, in a fashion, moved in step.
This a concern for more than speculators. Here's another way of looking at the importance of appropriate pricing.
Every 1 cent difference per bushel in corn prices swings costs for the ethanol industry and livestock sector by $50m each.
It has an impact on the price of daily bread too. For food companies, every cent swing in wheat prices represents more than $9m costs, potentially to be passed on to consumers.
That's why it's so important, for every consumer's sake, that the USDA gets it right. To be sure, the department is already an astonishing and invaluable resource - to the world - and the best in class by a light year.
But that doesn't mean it can't get better still.