Food Availability (Per Capita) Data System
Food availability doesn't equate directly to food consumption; spoilage and waste aren't accounted for. One of the data sets, Loss-Adjusted Food Availability, "adjusts aggregate food availability data for nonedible food parts and food lost through spoilage, plate waste, and other losses in the home and marketing system." It accounts for some loss but only goes back to 1970, that I can tell.
I can't determine how or if they accounted for food produced by a consumer, instead of by a marketing entity. That is, if you're eating eggs from your hen or carrots from your garden, are they accounted for in these data sets? It doesn't look like they are.
Still, as a proxy for consumption, they're revealing.
So, did we eat more meat 100 years ago than we do today? (Click graphs for larger.)
Fats and oils:
This one shows just oat products. What's that blip between 1988 to about 1995?
A couple beverage graphs ... First, coffee/tea/cocoa. Imagine if Starbucks was around in the 40s? (When I was growing up, everyone drank hot coffee with breakfast. I see a lot more carbonated beverages these days.)
Is America more a wine- or beer-drinking country?
Using data in their Loss-Adjusted Food Availability Set, I generated this graph for calorie consumption. We're eating (or what is available for us to eat) 519 more calories a day (24% more) than we did in 1970.
You can easily generate these charts on their site. The raw data is accessible in Excel worksheets too. It's what I used for the calorie graph. I could play with this all day.
Photo above is of a kitchen classroom in a housekeeping flat, New York, circa 1910, from Shorpy: The 100-Year-Old Photo Blog. Here's another: "December 1936. Farmer's wife churning butter. Emmet County, Iowa." (Click for larger.)