About a thousand new owners came through the door this year, most of them by way of the Blue Chip merger, so Buffett begins by telling them exactly what they have joined. Not a stock, in his telling, but a partnership that happens to be organized as a corporation.

The principles run to thirteen, and Exhibit 2 holds them all. The spine is one attitude: "Although our form is corporate, our attitude is partnership." Charlie Munger and he are the managing partners, and given the size of their holdings the controlling ones too. Every director is a major owner, most with over half the family's net worth in the stock. "We eat our own cooking," he writes, and few boards could say the same, or would want to.

Three rules do most of the work. Berkshire measures itself by progress per share, never by size. At a similar cost, it would rather buy $2 of earnings it cannot report than $1 it can, because whole businesses frequently sell for double the price of small pieces (last year's two pipes, now written down as policy). And every dollar of profit it keeps must in time become at least a dollar of market value, judged over rolling five-year stretches. So far, he reports, it has.

The rest are promises about temperament. Little debt, and long-term when any. No managerial "wish list" filled at the owners' expense. No selling a good business at any price, and no "gin rummy" habit of discarding the weakest business at each turn. Candor, with a warning attached: "the CEO who misleads others in public may eventually mislead himself in private." The one subject he will not be candid about is which stocks he is buying. Good ideas are rare, and easily stolen.

Here is the second choice in the flesh: pieces of businesses, bought through the insurers. The basket cost $566 million and was worth $1.31 billion at year-end, $2.31 for every dollar paid, with GEICO still the largest holding.

Crum & Forster, last year's parking place for cash, is gone from the list. Several positions were trimmed, while General Foods and R. J. Reynolds are far larger than a year ago. The forecast that matters is for the earnings nobody prints: Buffett expects these companies to mail Berkshire about $39 million of dividends in 1984 and to keep roughly $65 million more on its behalf. So far, he reports, the kept dollars have in aggregate become far more than a dollar each in market value.

The high point of 1983 was a furniture store. Berkshire bought 90% of Nebraska Furniture Mart from Rose Blumkin, who founded it with $500 and who, at 90, still sells carpet on the floor seven days a week.

Last year's letter quoted Pascal on the virtue of staying quietly in one room. This year Buffett leaves it: "Even Pascal would have left the room for Mrs. Blumkin." Her story is in Exhibit 5: out of Russia at 23 with no schooling and no English, a store opened in 1937 on $500, creditors paid "precisely as promised" by selling the furniture from her own home, and every Fair Trade case against her price-cutting won, one of them ending with the judge buying $1,400 of carpet.

The chain she built is in Exhibit 4. The Blumkins buy brilliantly, run the store at expense ratios "competitors don't even dream about," and pass much of the saving to customers, who reward one 200,000-square-foot store with over $100 million of sales a year, more than every Omaha competitor combined. Asked how he would compete with that, Buffett answers: "I'd rather wrestle grizzlies than compete with Mrs. B and her progeny."

Berkshire bought 90%, left 10% with the family members who run the store, and gave key young family managers an option on a further 10%. The store is not the only asset. "Geneticists should do handsprings over the Blumkin family," Buffett writes of Louie Blumkin and his three sons, and he ranks character ahead of buying skill. Berkshire would like more businesses of this kind, so the letter runs last year's want-ad again: repetition, as the newspapers tell their advertisers, is a key to results.

Book value per share rose 32% in 1983, and Buffett spends more words on what book value is not than on the gain. It is a record of money put in. What matters is what can be taken out.

He disowns the one-year number before anyone can admire it: "why should the time required for a planet to circle the sun synchronize precisely with the time required for business actions to pay off?" His yardstick is a five-year average, with red lights flashing if it falls much below what American industry earns as a whole. If that happens, watch out for the explanation, he warns, via Goethe: "When ideas fail, words come in very handy." Nineteen years of compounding have run at 22.6% a year, and his only forecast is that it cannot last.

Exhibit 6 draws why the two part company, with two cases from the letter. The textile mills earned nowhere near a fair return on the $19.46 a share put into them. See's $8 million of kitchens and stock earned $2 million a year. A business is worth the cash it will keep handing over, counted at today's prices, so the mills were worth less than was put in and See's about three times more. Buffett's own illustration is two children put through college at identical cost: identical book value, and futures worth anywhere from zero to many times the cost.

Berkshire has lived both halves. In 1965 its $19.46 of book value per share was all textile mills that could not earn a decent return, an investment he likens to a largely wasted education. By 1983 the sign has flipped: the business is worth considerably more than its books say, partly because of stock gains the books carry at cost, and mostly because of an asset they barely register. The next lesson is about it.

The hardest part, he admits, was unlearning. Taught 35 years earlier to prize tangible assets and shun businesses built on goodwill, he says the bias cost him many "mistakes of omission," and he borrows Keynes for the diagnosis: "The difficulty lies not in the new ideas but in escaping from the old ones."

Attached to the letter is an appendix on the strangest asset in accounting. Buffett says you can live a full and rewarding life without ever thinking about goodwill. He then shows why, in a world of rising prices, it is the asset that inflation feeds instead of starving.

Start with See's, bought by Blue Chip in 1972 for $25 million. Its tangible assets came to about $8 million, and it earned about $2 million a year after tax on them, a 25% return few businesses manage. The extra $17 million bought the thing producing that return: a reputation built on "countless pleasant experiences" with the product and the people. That is economic goodwill, drawn in Exhibit 7. Accounting goodwill is the same $17 million, written down by rule at $425,000 a year for forty years, whatever the reputation does. See's earned $13.7 million after tax in 1983 on about $20 million of tangible assets.

Then comes the experiment, drawn in Exhibit 8. Take See's and a "mundane" business that also earns $2 million but needs $18 million of plant and inventory to do it. Let prices double. To stand still, each must double its earnings and, in all likelihood, its tangible assets. The mundane business puts in $18 million and ends up worth only its assets, having gained a dollar for each dollar invested, like money in a savings account. See's puts in $8 million and, valued as before, gains about $25 million: over $3 per $1. "During inflation, Goodwill is the gift that keeps giving."

Two rules follow, and a warning. When judging how good a business is, ignore the amortization charge and look at what it earns on its tangible assets. When judging whether a purchase was wise, ignore the charge too, and count the full price paid, including the true value of any shares handed over. The warning is for goodwill of the other kind, created when "an overexcited management purchases a business at a silly price." The silliness has to go somewhere on the books, and this is where. Buffett's label for it: "No-Will."

The Buffalo News earned real money in 1983 after the losses of its Sunday-paper war, and Buffett uses it to show how the strength of a franchise can be read from outside: count how many households in town choose it every day.

For the six months to September 1983 the News ranked first in weekday penetration among the hundred largest American papers, and third on Sundays, "ten to twenty percentage points ahead of many well-known papers." It was not always so. Before 1977, Buffalo's only Sunday paper, the Courier-Express, lost readers in nearly every year from 1970 to 1976. The News now sells 376,000 copies of a Sunday paper that, before 1977, did not exist (Exhibit 12).

Three things explain the acceptance, in his telling: a stable population, an editorial reputation honed by the legendary Alfred Kirchhofer and carried on by Murray Light, and the simplest, drawn in Exhibit 11. The paper delivers an unusual amount of news. Half its content is editorial matter, against a sample of comparable papers in the high 30s. Some publishers have cut their news holes to lift margins. "We have maintained ours and will continue to do so." The News lives up to its name, which is more than most products can claim.

Not everything is cheerful. The year's profit margin just topped its 10% target with help from a one-time tax break and a drop in newsprint prices that will reverse, and advertising is drifting from the printed page into preprinted inserts, which earn less. Henry Urban, who "never flinched" through the dark days of the Sunday war, retired as publisher at midyear, and Stan Lipsey, with Berkshire since 1969, took over.

Buffett's own insurers had, in his words, "absolutely terrible results," while GEICO, which he does not run, "simply shot the lights out." He takes the blame for the first, explains why it took years to surface, and finds in the wreckage one kind of policy that is not a commodity after all.

The industry had its worst underwriting year in a long time, paying out about $1.11 for every premium dollar, exactly the kind of year last year's letter promised. Berkshire's own figure was far worse, $1.21, and Buffett declines the excuse that Mike Goldberg has only just taken charge: "the roots of the 1983 results are operating and personnel decisions made two or more years back when I had direct managerial responsibility for the insurance group." The same lag once worked for him. With GEICO, "we could picture what was likely to happen well before it actually occurred."

GEICO paid out about 96 cents per premium dollar, after setting aside a large voluntary sum for policyholder dividends, and Berkshire's roughly one-third share of its premiums, some $270 million, is now 80% larger than everything Berkshire's own insurers write. Jack Byrne and Bill Snyder keep the underwriting disciplined, Lou Simpson runs the investments, and Buffett's summary is that the three of them "are some team." None of it, he adds, lessens by an iota the need to fix his own operation.

The new business is the interesting part. Berkshire is pushing into reinsurance deals where "the buyer's overriding concern should be the seller's long-term creditworthiness." Structured settlements pay an injured claimant monthly, almost always for life, so the payments "must be unquestionably secure for decades to come." Other insurers also pay Berkshire lump sums to take over blocks of old claims. In both, the customer is buying decades of certainty, and no promise is a commodity when it has to last that long.

Shareholders keep asking why Berkshire will not split its stock to make a share affordable. Buffett's answer is that a split would change who owns the company, and that a busy stock market quietly picks its owners' pockets.

A rational share price needs rational shareholders, and Berkshire cannot screen applicants: "Mrs. Astor could select her 400, but anyone can buy any stock." What it can do is advertise honestly and let people sort themselves, the way an opera and a rock concert draw different crowds though anyone may buy a ticket (Exhibit 15). It is the self-selection the 1979 letter described with a restaurant, now with a box office. The result, he believes, is almost certainly the most owner-minded shareholder group among companies of its size, well over 90% of it held by people who owned Berkshire or Blue Chip five years ago.

A split would invite the opposite crowd, buyers who "feel wealthier with nine $10 bills than with one $100 bill," and "People who buy for non-value reasons are likely to sell for non-value reasons." At $1,300 a share, he adds, very few investors are actually priced out.

Then comes the arithmetic that gives the lesson its teeth (Exhibit 16). A company earns 12% on its owners' money. If every share changes hands once a year and each buyer and seller pays 1% in commissions (with the stock trading at book value), the owners as a group hand 2% of the company's net worth to brokers every year, a sixth of everything it earned, "for the privilege of transferring ownership." Hence his verdict on brokers who praise "liquidity": "what is good for the croupier is not good for the customer." Berkshire's own shares turn over about 3% a year, and its owners keep the difference.