There is a particular kind of shop sign that tells you a family expected to still be there in forty years. & Sons was not decoration. It was a claim about succession, made in public, on the letterhead. Somewhere in the last century American businesses stopped making it.
We went looking for the moment they stopped. The federal trademark register is one of the few naming records that runs continuously from the 1800s to now, so we loaded the full US Patent and Trademark Office case file dataset, 11.2 million marks, and counted how often each naming convention appears, decade by decade.
Every figure below is a rate rather than a count, because filings grew from 2,205 in the 1900s to 2,362,526 in the 2020s. A convention can triple in raw registrations over that period while collapsing as a share of everything filed. So the unit throughout is marks per 10,000 filed in that decade.
The fall
“& Sons” peaks in the earliest decade we will quote from and declines almost without interruption from there. By the 1960s it had fallen below one in ten thousand, and it has stayed there for sixty years.
That is a 39-fold fall. Among the earliest examples on file are names that are still recognisable: Steinway & Sons, E. R. Squibb & Sons, Abram Lyle & Sons, Hiram Walker & Sons. They read now as heritage brands, which is precisely the point. The form survives as inheritance, not as a choice anyone makes fresh.
The firms that did not survive are the more representative sample, and the register keeps their drawings too. The oldest “& Sons” artwork on file belongs to a London saddle soap maker who registered in the United States in 1890.
The related form, & Co, was far more common and fell harder in absolute terms. It held above 45 per 10,000 through the 1930s, then dropped to 10.6 in the 1940s and 5.5 in the 1950s.
Two decades either side of the Second World War account for most of a century of decline. We can see the shape clearly. We cannot tell you the cause from this dataset, and the honest position is that we have a striking correlation and no mechanism.
Firms kept filing in the old style well into that decline. A New York paper merchant registered this one in 1923, near the end of the period when a seal with the family name around the rim was simply how a business presented itself.
The form that never arrived
If “& Sons” is a claim about who inherits the business, the obvious question is what happened to the daughters.
The answer is bleaker than a slow decline. Written the way a firm writes itself, with an ampersand, the entire register holds 721 marks containing “& Sons” and 41 containing “& Daughters”. The earliest of those 41 was filed in 1992. This is not a decline from a peak. There was never a peak. For the first ninety years of the record the form does not occur at all.
It is worse than 41 on inspection. Most of those are not firms. They are slogans and programme names: Mothers & Daughters, Dads & Daughters, High Adventure Treks for Dads & Daughters. Filter to marks that are actually a family business naming itself and you are left with a handful. The one most people have heard of is Russ & Daughters, filed in 2002.
The convention that meant “this business passes to my children” ran for most of a century in a form that only admitted half of them, and by the time the other half appeared on a trademark application the convention itself was already dead.
Except for Brothers
One family form did not disappear. Bros and Brothers fell from 27.2 per 10,000 in the 1900s to about 4.6 by the 1960s, and then simply stopped falling. It sits between 4 and 7 per 10,000 in every decade since, including the most recent one at 5.6.
The reasonable reading is that “Bros” stopped being a statement of fact and became a style. Nobody reads a brewery called something Bros as a disclosure about the ownership structure. “& Sons” never made that transition, and the one that did is the one that survived.
That is not to say the firms behind the early ones fared any better. The Chicago meatpacker above is as gone as the saddle soap maker. The difference is that the naming convention outlived the businesses that used it.
What replaced it
The corporate vocabulary that arrived instead is easy to date, and it is more recent than it feels. Group is essentially absent before the 1960s, climbs to 38.6 per 10,000 by the 2000s, and has been falling since. Partners and Capital follow the same curve and peak in the same decade.
| Convention | Peak | At peak | 2020s |
|---|---|---|---|
| Group | 2000s | 38.6 | 27.6 |
| Capital | 2000s | 16.8 | 12.1 |
| Partners | 2000s | 14.0 | 8.8 |
| Ventures | 2020s | 4.2 | 4.2 |
| & Co | 1900s | 68.0 | 5.8 |
| & Sons | 1900s | 27.2 | 0.69 |
So the replacement has already peaked too. Only Ventures is still at its high, and only just.
The handover that is actually happening
The clearest signal in the modern half of the data is not a word at all. It is a legal form. Inc peaked in the 1980s at 117 per 10,000, the single most common convention of its era, and has fallen to 22.5. LLC went the other way from a standing start, and the 2020s are its highest decade on record at 34.3.
They cross in 2020, the first year in the register where more new marks describe themselves as an LLC than as incorporated. Both halves of that contribute: LLC climbed through the 1990s and 2000s, flattened for most of the 2010s, and resumed climbing, while Inc fell in almost every year after 2008.
Unlike the & Co collapse, this one is measured across millions of filings rather than hundreds, which is why the crossing year is worth stating to the year at all.
Which makes the long arc reasonably simple to state. A business used to name itself after the family that owned it. Then it named itself after the fact that it was a corporation. Now it names itself after the fact that it is an LLC, a form that did not exist in American law until Wyoming created it in 1977.
What this data cannot tell you
One hypothesis we tested did not survive, and it is worth reporting the failure.
Everyone knows the story about businesses called AAA Plumbing chasing the front of the phone book. If that were happening at scale you would expect marks beginning with A to bulge during the directory era and fade after the web killed it. They do not. Marks starting with A hold flat between 588 and 645 per 10,000 across the entire 150 years. AA and AAA peak in the 1950s and then sit near 5 indefinitely, long after the phone book stopped mattering.
We do not read that as the story being false. We read it as the wrong instrument. A plumber gaming alphabetical order registers a business name with his state and buys a directory listing. He has no particular reason to file a federal trademark. The register is a good record of what companies wanted to protect, and a poor record of what small local businesses called themselves. That limit applies to every figure above, not just to this one.
Method
Source: USPTO Trademark Case Files dataset, 2023 vintage, covering 11.2 million marks with text. Dates use the filing date where present and the registration date otherwise, because filing dates are sparse before 1900 while registration dates run back to 1870.
All rates are per 10,000 marks filed in that decade, and decades are cut on the year, so the 1940s means 1940 to 1949.
Claims start at the 1900s. The 1880s and 1890s hold 34 and 301 marks respectively, and a rate computed off 34 registrations is noise rather than a finding. “& Co” scores 588 per 10,000 in the 1880s on the strength of two filings. Those decades are excluded from every figure quoted here.
Marks are matched as filed and uppercased, on substrings with word separators enforced, so “Ventures” does not also collect Adventures and “Bros” does not collect Ambrosia.
This is a record of what businesses did call themselves, which is a different thing from what anyone should call one. We are not in a position to tell you the second, and the data does not support it anyway.