When it comes to recalls, just who is in charge of taking those food products off the shelves? It can be the retailer, it could be the state, but in most cases, it's not the federal government.

Recalls are on the public mind nowadays, and perhaps the greatest misconception about them is that the Food and Drug Administration orders them. The fact is that the agency has no legal authority to order recalls -- companies do so voluntarily.

If a company does not do so on its own, the FDA may request that it recall a product. The FDA has suggested guidelines on how companies should deal with recalls, and regional personnel sometimes meet with company officials to aid in designing recall strategy. In addition, the agency can pressure a company with publicity or pursue seizure of the product through the courts.

Nevertheless, most of the responsibility in conducting a recall lies with the recalling firm, states a 1984 General Accounting Office report that recommended to Congress that the FDA be given more authority to better protect the public from adulterated food products.

GAO's report included recommendations that the FDA should: (1) be given authority to detain products while processing seizure actions, (2) be given the authority to review manufacturers' records once adulterated products are identified, and (3) improve its monitoring of recalled food.

While voluntary recalls may spread the word faster because public awareness creates a setting for the process to move quickly, consumer groups have long advocated that the FDA be given more recall authority, said Ellen Haas, director of Public Voice for Food and Health Policy.

In fact, Sanford Miller, director of the agency's Center for Food Safety and Applied Nutrition, believes that the FDA should have more authority as well, but that members of Congress, who reflect the "emotional sentiment" of the country, are saying on the one hand, 'government protect us from everything,' and on the other hand are saying 'I don't want to give it government any powers because I don't trust it.' "

Another misconception about recalls is that there are only a handful of them per year. Varying in degrees of risk to public health and extent of distribution of the product, there were almost 1,000 food recalls in fiscal year 1985 -- ranging from "kerosene-like taste and odors" in a line of black-eyed peas to milk products contaminated with ammonia from a leaking refrigerator unit. Since Oct. 1, or the beginning of fiscal year 1986, there have already been 514 food recalls -- the majority of them due to undeclared sulfites or yellow dye number five on package labels.

Although the FDA publishes a list of recalls in its weekly enforcement report, it chooses to publicize only a handful of those -- the ones it believes pose a serious and immediate risk to a large segment of the public. Many recalls involve products in limited distribution.

In addition, the press may choose not to play up one recall or another. And sometimes too much of one kind of recall "loses impact," Miller said, the undeclared sulfite recalls being one example. Nevertheless, Miller said, the recalls were made an agency priority to "drive home the issue" within a diversified industry. (Miller also said that there are "some people who count the agency's effectiveness by the number of recalls and actions we take," but that the agency has no recall "quota." The FDA doesn't "play that game," he said.)

And often publicity about one recall will trigger public sensitivity and media attention about another. Miller said that after the Tylenol poisonings in 1982, there were several "me-too's" in which the public reported adulterated food.

The recent reports of glass in some Gerber baby food jars has been made much more complicated because it comes on top of the Tylenol poisoning, Miller said. Although the state of Maryland prohibited the sale of some Gerber products and area chains have removed some Gerber products until further notice, the FDA and Gerber have contended that a recall is unwarranted.

The recall process must begin, of course, with the discovery of an adulterated or contaminated food. According to Joseph Prendergast, the recall officer at the FDA's Center for Food Safety and Applied Nutrition, that may arise from a variety of sources such as a consumer, the manufacturer or a state health agency.

Early in February, two brie cheeses -- Joan of Arc with lot code number 323 and Esprit Nouveau with lot code number 351 -- both distributed by General Foods, were found to contain the bacterium listeria monocytogenes, which can cause flu-like symptoms in healthy individuals and stillbirths in pregnant women and be fatal to patients with cancer or AIDS.

The discovery was made quite accidentally, while FDA scientists in Cincinnati were fine-tuning the method for detecting the bacteria, said Peter Acly, spokesperson for General Foods. Using the company's imported brie as a test sample, the agency researchers stumbled upon the fact that it was contaminated, Acly said.

(Health agencies in southern California and the Centers for Disease Control had been investigating the listeria bacteria after medical personnel in Los Angeles began noticing an increase in the bacterial infection listeriosis, which was tracked to Hispanic women who had consumed Mexican-style cheese. The cheese was tracked to a California company which subsequently voluntarily recalled 44 of its dairy products and destroyed 1.5 million pounds of cheese. This outbreak prompted a nationwide FDA survey of soft-cheese manufacturers. In July, General Foods recalled all Liederkranz brand soft-ripened cheeses when tests for listeria proved positive. Late last month, the FDA also found the bacteria in several other brands of brie cheese made in Sorcy, France, the same place from which the General Foods cheese was imported.)

Acly said the next phase for the company was to conduct testing from other lots of the cheese. Two lots were identified, a hold was put on all cheese imported, and the company voluntarily issued a nationwide recall, not just of the two contaminated lots shipped to 24 states, but of all the cheese brands in question. Press releases were issued and consumers heard about the recall on the evening news.

Such recall announcements may take a while to filter through the system, and not always in an orderly fashion. Sometimes consumers find out about recalls before or at the same time that retailers do, posing a sticky situation for retailers who may be caught with the product on their shelves. Other times a company may issue a recall but it may be impossible to track all of the product because a subsidiary or distributor may have repackaged or incorporated it into another food (i.e., dried apricots containing undeclared sulfites into a trail mix).

The FDA sometimes doesn't hear about a recall until it is completed, but if it is handled well by the company that may not pose any problems.

Meanwhile, at the FDA last week, sitting on the desk of recall officer Prendergast was the General Foods brie file, containing interagency memos, lab reports and FDA district recommendations, plus statements from the agency's Health Hazard Evaluation Board. From all the available information, Prendergast will design a strategy for FDA follow-up and recall classification.

He has three choices for recall classification, according to the FDA:

Class I recalls are for products that could cause serious health problems or death. A recent example was a batch of frozen peeled shrimp imported by Red Lobster Inns of America that contained sulfites in excess of 100 parts per million. (None of the recalled shrimp was shipped to the Washington metropolitan area.)

Class II recalls are for products that might cause a temporary or reversible health hazard, or pose a slight threat of a serious nature. A recent example -- white hominy manufactured by Beatrice Specialty Products, enriched with iron and unsafe levels of niacin, according to the agency (recalls were initiated in Virginia as well as other southern states).

Class III recalls are for those products unlikely to cause any adverse health reactions, but which are in violation of FDA regulations. An example of this type of recall -- Sphite Co. Inc. in Roanoke, Va., had distributed imitation oil of cinnamon as pure cinnamon oil.

Miller said he doesn't always agree with the classification system and agreed that there are subjectivity problems in determining what is a serious health problem versus a slight threat of a serious nature. Sometimes the classifications create the wrong impression among the public and even among agency personnel, Miller said.

For instance, glass found in baby food jars would be considered a Class II recall, Miller said, if the concern was based on cuts on the lips or mouth. It's not unserious, Miller said, but "it's not the same thing as swallowing cyanide." (In fact, in November 1984, complaints of glass fragments in apple cherry and apple plum juices prompted Gerber to issue a recall of the juices in 17 states. It was classified as a Class II recall at the time.)

Then again, people with babies who cut their lips might say, " 'Why should that be a Class II?' If it were my baby, I'd be very upset" about that classification, too, Miller said.

Once the recall classification is made, a technician writes down the pertinent information in a notebook: the date the recall classification was made, the classification, the reason for recall and so on. In this age of computer memory banks, this has remained the agency's filing system.

Follow-up, destruction or reconditioning of the offensive food and monitoring the success of a recall is an area that may result in less than perfect results. In another GAO report on the FDA's oversight of the 1982 canned salmon recalls, GAO found evidence suggesting that two additional canneries that had produced defective cans should possibly have participated in the recalls. GAO also found instances of recalled products being mistakenly reintroduced into the market without being properly examined for defects.

Miller said that there can be "no such thing as a 100 percent effective recall." There are hundreds of thousands of retail establishments throughout the country and tens of thousands of warehouses. Anything can happen: "Some guy puts a recalled box in the back room and buries it -- accidentlly -- and then the clerk puts it back on the shelf," Miller said. Miller said that the agency is more than 90 percent effective in monitoring recalls, 95 percent effective with "really dangerous" problems. However, some adulterated food may slip through the cracks and be consumed, he said.

If you have questions or problems with food you believe to be adulterated, call the agency's Emergency Operations Branch in Rockville at 443-1240 or the FDA District Office in Baltimore at 301-962-3593.