Set back from Route 60 by a green field of winter wheat in the flat Texas panhandle country, 23 miles northeast of Amarillo, is the aging 10,000-acre World War II Pantex ordance plant.

It looks like the beef processing and packing plant on the same highway except for one detail -- Pantex areas are surrounded by double chainlink fences and modern guard towers.

Pantex clearly packs a different kind of product. In the middle of nowhere, it is the final assembly point for all U.S. nuclear weapons.

Over the next several months, specially constructed, unmarked, armored tractor-trailers, guarded by one or more armed escort vehicles will arrive at Pantex in increasing numbers from various parts of the country. Their cargo: nuclear and other special parts that will mark the beginning of the most ambitious U.S. nuclear weapons building program in 20 years.

The program comes at a time when the United States and the Soviet Union are engaged in the final negotiations on a strategic arms limitation treaty, which is a major goal of the leaders in both countries.

Over the next five years, the nation's nuclear stockpile is to receive new MIRV warheads for Minuteman III land and sea-based intercontinental missiles; a new cruise missile; a new Lance warhead; new strategic and tactical bombs, and a whole family of atomic artillery shells.

The new weapons are more powerful and effective and, according to scientists who worked on them, safer from accidental explosion than those they will replace.

A government scientist said recently that "a nuclear weapon these days is like a complex pinball machin."

There are, for example, over 2,000 separate parts in the new Trident I submarine-launched ballistic missile warhead. Some 125 separate components will come together at Pantex when the Trident I production gets underway in the coming year.

The process by which the United States develops, tests and approves production of nuclear weapons such as the Trident I is as unknown to the public as the Pantex facility and the rest of the multibillion-dollar nuclear weapons complex. It is a network of government-owned laboratories, testing facilities and manufacturing plants, located, for security reasons, unobtrusively from California to Florida.

Starting with the World War II Manhattan Project, which led to the first atomic bombs, secrecy has been the hallmark of the U.S. nuclear weapons program.

That secrecy has relaxed somewhat in recent years. But what the public has learned primarily is about numbers -- the number of weapons, the total yield or the power of each device. These types of numbers baffle but seem to satisfy the public and keep people away from the realities of the nuclear weapons building process.

The result has been that real knowledge and decision-making in the nuclear weapons field has been held by a relatively small group of government technocrats and scentists, both civilian and military.

Like the high priests of old, they have become keepers of the flame.

By law the president each year must review and approve nuclear weapons tests as well as the production, deployment and retirement of all weapons.

This "stockpile" paper which he reviews usually arrives in November or December of each year.

The stockpile paper is normally drawn up by the Department of Energy -- which builds nuclear weapons -- in consultation with the Defense Department, the consumer of the weapons.

As consumer, the Pentagon must decide what type of weapons it wants. As in the civilian marketplace, however, the consumer is limited by what is offered. In the case of nuclear weapons, the nation's two nuclear weapons laboratories at Los Alamos, N.M., and Livermore, Calif., often determine characteristics for a weapon because of what they have designed beforehand.

Personnel, too, overlap in the nuclear weapons field, with a small number of technocrats and military men shuffling from post to post.

ODE's past chief of national security operations was an Air Force general.

His replacement, the present DOE assistant secretary in charge of weapons, is the former deputy director of the Livermore weapons laboratory. His chief deputy at Doe/ worked at Los Alamos. The DOE deputy in charge of military applications is an Army general.

Over at the Pentagon,the man who runs the Military Liaison Committee (MLC), which coordinates between Defense an DOE, is a former Livermore man. And, for the ifrst time, the secretary of defense, Harold Brown, is himself a former Livermore director.

The clique has also moved in on Capitol Hill. The men who staffed the House and Senate Armed Services committees, overseers of the nuclear weapons complex in the last session of Congress, were themselves both veterans of this complex. One, Seymour Schwiller of the House committee, once worked at Los Alamos and actually was a DOE employe on loan to the committee. Donald C. Cotte, who recently left the Senate committee job, previously had been head of the Pentagon's MLC.

Although the group that runs the nuclear complex is narrow, the system that leads to a decision to produce a new warhead seems rational and secure -- at least on paper.

But interviews ove the past few months with those who have participated in the process -- present and former government officials, military men and scientists, members of Congress -- disclose a somewhat different story.

The process, according to many, has been mindless at times, grinding out new weapons to replace old without critical aualysis of how they eventually could be used.

Nuclear artillery is a case in point. The first nuclear shells were built in the early 1950s when the United States was concerned about a European war. All weapons were made with dual-capability for nuclear and conventional warheads.

Twenty years later, the limits of nuclear artillery in range and use are well-known. Yet a new generation is about to be produced without a close look at whether they are in fact needed.

Inter-service rivalries also have affected nuclear weapons decisions. The Army's desire to get into the nuclear picture, for example, was as much behind the push for nuclear artillery as any other factor.

In recent years, competition between Los Alamos and Livermore has been a growing influence on the weapons program as they try to outbid each other to produce a warhead.

Enhanced radiation effects from nuclear weapons, the basis for neutron warheads, is a Livermore product. It was promoteda as the basis for a new 8-inch nuclear artillery shell when Congress in 1973 turned down funding Los Alamos' traditional nuclear shell.

Today the two laboratories are working on competitive medium-range missiles for Europe and each may have a special weapon effect as asales point.

The bureaucratic framework for the competition is DOE's seven-phase development and production system for nuclear weapons. The complicated route from conception to production normally takes from five to 10 years. Stockpile life for a completed warhead is supposed to be 15 to 50 years.

Phase one is weapon conception, where the two nuclear laboratories explore on paper ideas that oculd be translated into deliverable systems.

In some phase one studies, the Los Alamos and Livermore labs coordinate with the Sandia Corp., which designs electronic systems. These miniaturized systems inside the weapon guide it to the target, prepare it for firing, and then actually set it off.

Phase two involves more study, often using computers.

The Los Alamos and Livermore weapons labs have two of the most advanced computer complexes in the country. They use the computers to analyze through mathematical models exactly how the nuclear elements in the proposed weapon would work and whether the outcome is what the military services need.

Scientists at both weapons labs insist these computer exercises are the single most important tool in the development process. Without the mathematical projections, there would be no way, short of exploding hundreds of test devices, to determine the size and shape of new weapons.

It is also in this phase that some devices are tested underground in Nevada to check out what the computers have shown to be the most promising designs. Both labs have an allocated number of such tests -- seven to 12 in recent years -- which can cost $1 million or more a test.

The mumber of development tests allowed by the president has decreased in past years. With the possibility of a comprehensive test ban looming for the future, both labs have been pressing to increase this number.

Phase two has one other important element. The military services by then have shown an interest in the type of weapon under study. The prospect exists of a warhead emerging that will go on to production. Bouth labs complete with separate designs -- a competition that ends with one or the other being selected by DOE to develop the weapon if it is chosen for production.

These first two phases can run three years or more. Many weapons studies never proceed beyond phase two. Currently, for example, the Navy has made a request for a phase two study of a nuclear antisubmarine warfare weapon. "They ask for that every few years," one lab official said recently, "but we doubt it will ever go anywhere."

Phase three is the jump to committing the large amounts of funds necessary to build a weapon.

The Pentagon must request a phase three study. DOE officials then decide whether Livermore's or Los Alamos' design will be used. This decides where engineering and subsequent development will take place.

It is normally during phase three that Congress, through budget submissions, gets its first indication that a specific new weapon is being contemplated.

It is also during phase three that the first sign of the costly, intricate and tiem-consuming effort that goes into manufacturing each weapon emerges. For example, DOE's Albuquerque Operations Office, which coordinates the production process, produces a three-volume plan assigning production of warhead parts to different DOE plants.

Thus, during phase three, the first orders will be placed for the special machine tools to build the unique parts that each weapon needs. The lab designing the weapon sends out specifications, and each plant in the complex works out a production system. Although the labs can fabricate one prototype of the weapons, other plants in the complex have to make their particular specialty for the eventual production run.

Frequently not only are new machine tools needed to build the parts designed for a specific weapon, but also a machine must be built to test the new part to ensure it meets specification.

Some of these pleces of equipment are so complex that they must be ordered more than three years before production is scheduled to begin.

Late in phase three, th e proposed warhead finally is priced, not just in terms of nuclear and other production materials, but also in terms of machinery and actual construction needed to meet its unique requirements.

Those additional costs are often substantial. For example an extra $22 million was programmed for construction associated with the proposed neutron 8-inch artillery shell.

Phase three can take two to three years or more to complete.

Phase four -- final production engineering -- lasts another two years. During this time a work force to build a weapon is security-cleared and trained.

Working models of the warhead are built. Tests of parts are made constantly -- pushing the material until it is destroyed to see how long it lasts. Hundreds of changes are made as a result of these tests, with the labs that designed the weapon working with the production personnel.

Working out the kinks in production during phase four can take another two years or more.

Pilot production begins in phase five, with the parts going to the Pantex plant in Amarillo for assembly. Here again, completed weapons are then torn part to see how various parts interacted.

Nuclear weapons are made up of materials that are noncompatible. The rock-hard high explosive components are often chemically unstable, while the metallic nuclear materials are constantly giving off radiation.

The glues, resins and epoxies, can react to the chemical and nuclear materials tightly packed inside the warhead container, That reaction can harm the complicated miniature switches, wires and other electronic gadgetry.

Production is also slowed down by limitations on how much nuclear and high explosive material can be in an assembly area at one time.

Normally, less than a year after pilot production began, phase six-full-scale stockpile production -- starts. The time lapse is frequently 10 years since the process began.

From Pantex, the completed warheads -- sometimes mated to a delivery system -- are shipped in guarded armored trains to the military customers.

Dilivery to the stockpile does not end DOE's responsibilities for the warheads. Each year a number of each weapon are returned to Pantex, the nuclear companents removed, the weapon checked and delivery systems tested.

When the time comes for retirement -- often after 20 years -- disassembly also takes place at Pantex.

In dollars alone, the nuclear warhead operation is far from big defense business. About $1.4 billion a year is spent on weapons development and production, including the costs of the nuclear weapons labs. Another $900 million goes for nuclear materials such as plutonium, tritium, and enriched uranium.

Though the dollars aren't big, some of the biggest corporations in the country operate the major weapons building plants.

Western Electric's Bell Laboratories, for example, runs Sandia Corp., which designs firing and fusing components near the two nuclear labs.

The DuPont Co. operates the Savannah River, S.C., facilities that produce plutonium and tritium.

Union Carbide runs Oak Ridge, Tenn. plants that reprocess enriched uranium for use in weapons. No new enriched uranium has been made since 1962. All the material for new weapons comes from those weapons that have been retired.

From South Carolina, plutonium is trucked across the country in heavily guarded armored tractor-trailers to Rocky Flats, Colo. There, Rockwell International runs the facility that fabricates the plutonium for use in weapons...

Three other plants around the country play roles in the creation of each weapon.

At Kansas City, in a plant operated by the Bendix Corp., the electrical and electromechanical parts of each weapon are built. In Miamisburg, Ohio, Monsanto Corp. manufactures detonators for the weapons, In Pinellas County Fla., General Electric Co. produces neutron generators, needed to promote the nuclear reaction.

Amarillo's Pantex Plant is operated by Mason and Hanger-Silas Mason Co., Inc., a firm that has been in the chemical high explosives business for more than 100 years. In addition to serving as the final assembly point, the Pantex plant also fabricates the high explosive components of the weapons.

Past history shows that problems come up regularly with the stockpiled weapons. A frozen safety switch in the mid-1960s would have kept three-quarters of the Polaris A1 warheads from exploding with nuclear force had they been launched.

Building nuclear warheads is like no othr manufacturing operation in the United States. From scientist ot plant general manager, the participants express great satisfaction that their products are like no others. The weapons probably will never be used. They must, however, sit ready for immediate use for 20 years. The men that design and build them say they are expected to work with 98 percent reliability.

But it is a claim that no one expects to be around to check up on.

NEXT: Tactical nuclear arms