So the phone rings at a big publishing company in New York. “How long is a book? ” asks the caller.

“Well it varies from book to book and genre to genre,” explained the publishing company receptionist.

“This is a novel. How long is a novel?” the caller asked.

“That varies, too, but many of ours are around 80,000 words,” the receptionist said.

“Thank God, I’m finally finished!” said the caller.

By the same token, how much money does it take to start a technology business? I’ve just spent the summer with more than 30 startups and can tell you the amount varies greatly — more than you could even imagine.

In the simplest sense how much money it takes to start a technology business depends mainly on how much money you have, because it generally takes it all. But all can vary a lot.

The most money raised by any of the Startup Tour companies we visited this summer was $70 million. The least was $5. There were plenty in the $1+ million range but I’d guess the median was around $40,000.

There are plenty of companies that claimed to have not raised any money at all, but that’s not true. The founders of those companies generally went without pay for six months or more, so their companies were self-funded with significant dollars. That makes the $5 company all the more amazing, because it really did start with just $5 — for business cards at Kinkos — and was profitable before the end of its first day in business.

Remember these are companies outside Silicon Valley. Most of the companies we visited this summer had never even met a venture capitalist. Most were funded by family and friends. A surprising number relied on government funding, primarily in the form of Small Business Innovation Research (SBIR) grants.

If there’s a role for government in encouraging tech startups, SBIR defines that role. For those unfamiliar with the SBIR program, federal agencies that spend more than $500 million per year on outside research are required to set aside a small percentage (I think it is two percent) of that money for research contracts with small businesses. The two-phase contracts are for $100,000 and $600,000 to develop technologies of interest to the government. But while the government gets use of the technology, they don’t get to own it or even have equity in the developing company, so from an entrepreneurial standpoint this is ideal.

From what I have seen, the SBIR program is modest, yet extremely successful at encouraging innovation. Perhaps it should be expanded.

Yet that’s about as far as federal success in this area goes. None of the startups had done business, for example, with the Small Business Administration or with SBA lenders. For technology at least, this more traditional program is a non-starter, probably because it is hard to explain to a bank the value of software.

A lot of what we looked at was software, but home equity loans also funded a robotics company and a solar company and probably other companies we didn’t even realize were built on housing bubble money.

The point is that it doesn’t take a lot of money — certainly not Silicon Valley-type money — to start a very fine company. The trick is to either do-it-yourself or do-it-offshore, with the offshore model oddly in decline, probably since there are so many out-of-work engineers in the USA.

One of the more surprising conclusions of the summer is that many of our companies saved so much time by not looking for money that they ended-up not needing the money they might have raised.

Let me explain this last point in more depth. If you spend three months writing business plans and visiting VCs before you have a prototype, then you are three months late (and $X behind) before the first line of code is written or first piece of metal cut. Yet you had to eat during those same three months. Better to go for 90 days on savings or on a 30-second pitch to your rich uncle than to waste three months looking for VC money.

Get a good prototype and the money may come looking for you.

And certainly six months is enough time to know if your idea is going to work or not. So six months of income is the most you should expect to raise or spend from savings.

I don’t care if you are inventing a frigging immortality drug, the same funding rules apply, at least outside Sand Hill Road.

We visited a very promising pharma startup, for example, that had so far spent only $30,000. Yes, a lot more money would be shortly needed for large animal and human trials, but the preliminary work was done, their basic IP was protected, and no equity was burned in the process.

It’s painful for them now, but in the end each of these companies will be glad they were so careful with their spending.

Not that they all were so careful. Home equity money circa 2006 was so abundant many of our founders made stupid mistakes. But to make our list they also recovered from those mistakes. And when you look at the dollars that actually bought the right stuff, they generally came down to that same $30-40K.

So how much money does it take to start a technology company? Less than you think. Maybe even less than you have.