One of the major reasons Virginia Republicans have always said they wanted to run John N. Dalton against Henry E. Howell for governor was that they believed they knew Howell and his weaknesses.

After all, it's hard to have been a fixture on the state's political scene for more than two decades and be an unknown. And Howell, most Republicans have believed, was, for all his seeming lack of orthodoxy, a known quantity.

Now, on the heels of Howell's upset victory over former state attorney general Andrew P. Miller, comes an unsettling suggestion that Howell may not provide the Republicans the familiar target they were expecting.

That, at least, is one of the worries a recently released poll by three College of William and Mary government professors is giving some Republicans.

While both Dalton and Howell strategists have quibbled over the poll's reliability, neither camp is questioning its central, most-publicized findings: That the race this fall is likely to be close; that both Howell and Dalton have gained support since the June 14 Democratic primary and that the larger the vote this fall, the better off Howell will be.

The poll, the product of interviews taken in the Williamsburg area before and after the primary, confirms the finding of a Washington Post primary day survey that about seven out of ten Miller supporters are leaning toward supporting Dalton in November.

But what probably is one of the poll's most significant findings has been overlooked in most public accounts. That is the suggestion that Howell succeeded during the primary in reversing his image from that of a free-spending liberal to that of a close-fisted conservative.

One of the key elements of Howell's campaign was his promise that he would not support any new general tax increases during his four years as governor. Miller never took the Howell statement seriously, saying that it was impossible to predict what would happen in a four-year period.

Voters, however, appear to have taken Howell at his word, according to the Williamsburg survey. Among voters expressing an opinion in the survey, Howell was initially perceived as a man more likely to raise taxes than to cut services in order to reduce the costs of government. The margin was admittedly small - 19 per cent thought Howell would cut services and 21 per cent thought he would favor higher taxes.

But after the primary, the figures were reversed. In the words of William and Mary's John McGlennon, an assistant professor of government, "something was starting to emerge." By a margin of 26 to 19 per cent, those surveyed believed Howell was more likely to cut services than to raise taxes.

In short, by the end of the primary, Howell, "among those who did have an opinion, began to emerge as an anti-tax candidate," McGlennon says.

At the same time, Howell was improving his standing among the electorate as a man believed to be "very" or "fairly honest" and his recognition as an opponent of recent utility rate increases.

Miller, by comparison, "seemed to make very little impact at all" on the voters, McGlennon says. "Of the three (Howell, Miller and Dalton) he showed the least change" during the primary. McGlennon admits this finding seems to be "amazing" in view of Miller's record $1 million primary spending.

Publicly, Republicans downgrade the Williamsburg survey, saying it was not done in accordance with professional standards and that the Williamsburg area is hardly a mirror of the Virginia electorate. "We don't make any claim that it is very reflective of the entire state," McGlennon counters.

He quickly adds, nonetheless, that voters in the area surveyed came within one percentage point of the statewide 1973 Virginia gubernatorial results and last year matched exactly Gerald Ford's 51 to 49 percentage victory over Jimmy Carter in the state.

Despite his misgivings over the poll. William A. Royall, Dalton's manager, is among the first to congratulate Howell for his skillful primary victory and his apparent ability to reverse his public image on government spending. "There is no question he was successful about fooling the voters on that," Royal said recently.

Projecting the conservative Howell image to the general electorate. Royall argues, will be more difficult this fall because Republicans are not going to make the same mistakes that Miller made. Dalton already seems to be at work on that theme.

During a recent campaign stop in Northern Virginia, he made a point of saying Howell hasn't changed at all. "He's the same old Henry." Dalton remarked in a radio broadcast. "They may have dressed him up in three-piece business suits and he may talk more moderate these days, but he's the same man who introduced a bill to abolish Virginia's right-to-work law and voted for George McGovern in 1972," Dalton said.

That is, of course, the Howell that Gov. Mills E. Godwin defeated in 1973 and the man who lost in a 1969 Democratic primary for governor. And it is the Howell that Dalton wants to run against.

But as even Royall will concede, it is probably not the Howell that Dalton will find himself facing this fall.

Republicans undoubtedly will argue that Howell has no claim to a conservative image and that Dalton, who has been espousing a strong conservative theme the past four years, is, as Royall puts it, "closer to where the average Virginian finds himself."

That assumes Virginians will not support a candidate whose positions have changed over the years and that Virginians are, in fact, as conservative as their reputation. Howell claims that the economic times have changed and that some programs he once advocated are no longer feasible.

Furthermore, McGlennon says those questioned in his survey "don't come across as rigidly conservative." True, more opposed collective bargaining for public employees, as Dalton does, than support the concept, as Howell does.

But more also said they would support a call for higher taxes rather than face an across-the-board cut in state services (a 35 to 29 per cent margin). An overwhelming percentage (53 to 18 per cent) said they would favor strong environmental controls instead of allowing unrestricted industrial growth.

Just how accurate a barometer the Williamsburg poll is won't be known until Nov. 8. Until then McGlennon adds that he, Alan Abramowitz, and Ronald Rapoport, the professors who oversaw the project, expect more Howell and Dalton staffers to continue drifting into Williamsburg to glance at the latest analyses their computer has developed.