In 1977 Berkshire made about $22 million in profit from running its businesses, and the profit for each slice of the company jumped 37% over the year before. That sounds great. Buffett's first job is to explain why, on its own, it isn't.

Yes, profit per slice rose 37%. But the owners had also put 24% more money into the business that year, so most of that jump was simply a bigger pile of money making more money, not the company getting smarter. The figure Buffett actually cared about was this: for every $100 of the owners' money the company began the year with, it earned about $19 of profit. A 19% return, better than most companies manage, and far better than leaving cash in a bank.

Berkshire's original business was making cloth. In 1977 it had another bad year, and Buffett had wrongly promised it would improve two years running. That, he admitted, might say something about the textile business, his own forecasting, or both. He uses it to make an honest point about effort.

The cloth business kept struggling with both making and selling its product. Buffett kept it going anyway, and he is honest about why: the factories were among the biggest employers in their towns, the workers were loyal and skilled, and years earlier this same business had quietly supplied the cash that built Berkshire's far better insurance arm. But he does not pretend. In cloth-making, he says, even very good management can earn only modest results.

Insurance was the opposite story, and Berkshire's big winner. An insurer collects money from many customers up front and pays out only to the few who later suffer a loss. Done well, it is a wonderful business. Done badly, it is a wonderful business for the customers.

When Berkshire entered insurance in 1967, its insurers collected about $22 million a year from customers. By 1977 that was $151 million, roughly seven times more, and Berkshire never had to sell new slices of itself to pay for the growth. Why was 1977 so strong? Insurers had pushed through big price rises in 1976 after some dreadful years, and those higher prices only fully landed in 1977. Even a few honest mistakes could not spoil a year like that.

Buffett adds a warning. The good times were already fading. The cost of paying claims was creeping up about 1% every month (things get pricier over time, and juries keep deciding insurers owe more). Unless prices kept climbing too, profits would shrink, and he expected them to start shrinking later in the year.

Insurance, Buffett points out, is almost identical from one company to the next. Everyone sells the same thing: a promise to pay if something goes wrong. Anyone can get a license, and the prices are public. Nothing makes one insurer special except how well it is run, which is why he praises his managers by name, like a coach naming star players.

Berkshire's insurers held a big pile of cash, and Buffett invested it by buying shares of other companies. The trick: he buys a few shares thinking exactly like someone buying the entire business.

Here is what Berkshire paid for its biggest shareholdings, beside what those shares were worth at the end of 1977. The first thing worth noticing is not the gains, but when Buffett bought.

The Washington Post stands out: about $10.6 million paid, worth roughly $33.4 million. Yet Buffett warns against getting carried away by a number like that. He reminds shareholders that any single date's figure is unreliable: at the end of 1974 his stockholdings carried an unrealized loss of about $17 million, only three years before they showed a $74 million gain at the end of 1977. He judges his choices by how the businesses perform over many years, not by the price on any single day.

Buffett tells an embarrassing story about his own company. Back in 1948, Berkshire's cloth mills were a powerhouse: about $18 million in profit, 10,000 workers, profits in the league of giants like IBM. A snapshot from 1948 would have screamed "blue-chip winner."

Buffett's favorite kind of business makes plenty of profit without needing much new money to keep growing. These rarely have a flashy year, which is exactly why people overlook them.

Berkshire owned a small bank in Illinois. For its size it earned about three times as much on its assets as most big banks, while staying very safe. It was started in 1931 with $250,000 and earned just $8,782 in its first full year. No new money has gone in since; instead, over the years, it has handed its owner $20 million in cash, a bank that gives money back rather than asking for more. It earned $3.6 million in 1977.

Berkshire also owned a slice of a candy company, See's Candies. In just five years its yearly profit grew from $4.2 million to $12.6 million, roughly tripling, and it managed this with very little new money, in a candy market that was barely growing at all. People simply loved it enough to pay more.