The nation's middle class is "holding its own" and is not declining as a proportion of the work force, despite popular theories that it is shriveling up, according to a new study published in the government's Monthly Labor Review.

The author, Neal H. Rosenthal, an official of the Labor Department's Bureau of Labor Statistics, studied the theory that an increase in the proportion of service jobs and a decline in the proportion of well-paying manufacturing jobs is grinding down the middle class between ever-swelling lower and upper classes.

He concluded that it isn't so and is not likely to happen in the future either.

Rosenthal studied changes in the distribution of the labor force and in relative pay scales among 416 different occupations from 1973 to 1982.

He concluded that the top one-third of occupations, ranked by pay, had proportionately more workers in 1982 than in 1973, the middle third had about the same in 1982 as in 1973 by one method of analysis and a larger share by another, and the lowest one-third had fewer workers.

"These results do not bear out the notion" that proportion of workers in the middle-paying one-third of occupations is shrinking and leading to bipolarization of the labor force, Rosenthal said.

Moreover, he concluded that future developments, such as the decline of well-paying "smokestack" industries and the increase in service and high-technology jobs, on balance would not lead to bipolarization.

In one part of his study, Rosenthal ranked the 416 occupations on the basis of their normal weekly earnings in 1973 and in 1982.

He found that in 1973, the one-third of occupations with the highest normal pay had 27.7 percent of all full-time workers, the middle third had 28.9 percent and the bottom third had 43.4 percent. In 1982, however, the top third of occupations had increased to 29 percent of all full-time workers, the middle third 33.4 percent and the bottom third declined to 37.6 percent.

Including part-time workers as well, he found that the share of workers in the highest-paying one-third of occupations rose from 22.8 percent to 24.8 percent from 1973 to 1982, the middle-third was virtually unchanged, declining from 31.1 percent to 30.5 percent, and the lowest third declined from 46 percent to 44.6 percent.

He said use of the calculation including only full-time workers probably is more meaningful since many part-time workers are students or part of households having a wage-earner with a high income, and the part-timer's wages have little long-range bearing on earnings-class differentations.

He also found that average weekly earnings in each of the three categories increased by about the same percentage from 1973 to 1982.

"These results do not support the notion of bipolarization," said Rosenthal. He said none of the analyses showed an increasing share of workers engaged in the lowest-paying one-third of occupations, "which is an important part of the bipolarization hypothesis. In fact, they all show a decline in the share of employment of the lowest group."

Looking at the future, Rosenthal concluded that while some factors such as the decline in the number of "smokestack" industry jobs could lead to a decline in the middle class, other projected developments offset them.

For one thing, he said, "workers who typically have a high level of earnings -- professional and technical workers and managers -- are projected to increase as a proportion of all workers from 1982-95." They will go up from 16.3 percent of total employment to 17.1 percent.

And while the proportion of jobs in well-paying "smokestack" industries like auto and steel will go down, the proportion in some other industries that are low-paying will also go down (textile, apparel and leather product manufacturing).

In addition, he said, pay of production workers in high-tech industries is generally above that for production workers in manufacturing.