dissenting.
A union in dispute with a grocery bag manufacturer, in order to induce consumers to boycott the bags, put up pickets at some Kroger stores where the bags were used to sack groceries at the counter. The pickets asked the customers not to use the bags. The NLRB found that the picketing did not and could not cause serious economic harm to the stores and ruled against Kroger’s claim that the picketing constitutes an impermissible secondary boycott, an unfair labor practice under § 8(b)(4)(ii) of the Act.
I find the opinion of our Court on these facts unconvincing in light of the reasoning *640of the Supreme Court in its recent secondary boycott opinion in the Safeco case, 447 U.S. 607, 100 S.Ct. 2372, 65 L.Ed.2d 377 (1980), adhering to and refining the Tree Fruits distinction between “enterprise” and “product” secondary picketing. Secondary picketing directed at the enterprise is illegal, but product picketing is illegal only under narrow circumstances: when it threatens the secondary employer with severe loss.
In Safeco, the neutral employer sold only one product. The Supreme Court, therefore, said that “the application” of the statute “to [prohibit] the picketing in this case” does not violate the First Amendment. This statement immediately follows footnote 11 in which the Court states the test to be applied in a case such as ours, a test designed to define a category of harmful product picketing that clearly falls outside the protection of the First Amendment:
The picketing in Tree Fruits [apples at a neutral grocery store] and the picketing in this case [a particular brand of insurance at a neutral one-product insurance agency] are relatively extreme examples of the spectrum of conduct that the Board and the courts will encounter .... If secondary picketing were directed against a product representing a major portion of a neutral’s business, but significantly less than that represented by a single dominant product, neither Tree Fruits nor today’s decision necessarily would control. The critical question would be whether, by encouraging customers to reject the struck product, the secondary appeal is reasonably likely to threaten the neutral party with ruin or substantial loss. Resolution of the question in each case will be entrusted to the Board's expertise. 447 U.S. at 613, 100 S.Ct. at 2376 (emphasis added).
The essential points here are (1) the statement that the test under § 8(b)(4)(ii) is the reasonable likelihood of threatened “ruin or substantial loss” and (2) the statement that this factual finding is “entrusted” to the NLRB.
In addition, five members of the Court in Safeco — Justices Blackmun and Stevens, who wrote separate concurrences, and Justices Brennan, White and Marshall, who dissented — appear to believe even more strongly that secondary boycott cases such as this one in which the restriction on the picketers’ speech is content-based raise “difficult First Amendment issues” (Blackmun, J., concurring), requiring the Court “to determine whether the method or manner of expression, considered in context, justifies the particular restriction” (Stevens, J., concurring).
The First Amendment problem inherent in any governmental prohibition of picketing based on the content of the message is what led the Supreme Court initially to adopt this construction of § 8(b)(4)(ii) in Tree Fruits distinguishing between enterprise and product picketing. In that case, Justice Brennan wrote for the Court that the reason for reconstructing the statute to include this distinction — an otherwise somewhat strained construction — was the Court’s “concern that a broad ban against peaceful picketing might collide with the guarantees of the First Amendment.” 377 U.S. at 63, 84 S.Ct. at 1066. Justice Black believed the statute unconstitutional because he was unwilling to reconstruct the statute in this way in light of what he conceived to be a clear legislative purpose to broadly ban all secondary picketing. 377 U.S. at 76-80, 84 S.Ct. at 1073-1075. Thus the distinction and the Supreme Court’s refinement of it in Safeco are clearly based on First Amendment imperatives.
In this case I do not believe that the Board went beyond the range of administrative discretion “entrusted” to it in finding that this boycott of grocery bags would not cause Kroger substantial economic harm — whether the boycott were successful or not. The record is devoid of any facts suggesting that the boycott of the Kroger stores was “reasonably likely to threaten [Kroger] with ruin or substantial loss.” In fact, the record is clear that the boycott had no effect at all on Kroger’s business because consumers rejected the idea the boycott promoted.
*641It is true that we should not judge the legality of the boycott, the particular protest in question, entirely on the basis of whether the idea is accepted or rejected by the consumer. But I do not think our Court is correct in ruling out altogether any consideration of this factor and requiring us to assume — contrary to fact — that the boycott will be successful.
Our Court’s conclusion that we must assume that all boycotts will be successful appears to violate First Amendment principles. In cases such as this where the prohibition against expression is based in part on content, the state must show a clear danger or compelling justification. Causing “ruin or substantial loss” to the neutral employer is such a justification. Our Court assumes that the “danger” of any secondary picketing will always be “clear and present,” that the “justification” will always be “compelling.” But in order for the expression to fall outside the protection of the First Amendment, it must not only be “directed to inciting or producing imminent lawless action” (ruin or substantial loss) it must also be “likely to incite or produce such action.” Brandenburg v. Ohio, 395 U.S. 444, 447, 89 S.Ct. 1827, 1829, 23 L.Ed.2d 430 (1969). Not only must the threatened harm be severe if it should occur; it must be likely to occur, not remote or just a possibility. Thus the harmless inciter, or the harmless labor picketer, is protected. See Brandeis, J. concurring in Whitney v. California, 274 U.S. 357, 377-78, 47 S.Ct. 641, 649, 71 L.Ed. 1095 (1927), overruled by Brandenburg. The facts in this case show that the picketing did not in fact, and was not “likely to produce,” harm to the neutral employer.1
More recently, in Police Department of Chicago v. Mosley, 408 U.S. 92, 92 S.Ct. 2286, 33 L.Ed.2d 212 (1972), and Carey v. Brown, 447 U.S. 455, 100 S.Ct. 2286, 65 L.Ed.2d 263 (1980), the Supreme Court struck down under equal treatment or content-neutrality principles of the First Amendment two statutes that prohibited picketing in front of schools and in front of homes, except that in both cases the statutes made an exception in the case of labor picketing. Labor picketing, but not other picketing, was allowed. Quiet residential streets, serene homes, and undisturbed classrooms were not thought sufficient state justification for a statute discriminating in favor of labor picketing. Surely then the negligible economic impact shown in this case is not sufficient justification for an application of a statute which discriminates against a particular kind of labor picketing on the basis of the product in question.
Here not only was the “danger” remote, the “justification” uncompelling. There was no danger in the theoretical sense, no justification for the state’s interference with the citizen’s liberty of expression because there was no possible harm. Even if we assume success, even if we assume that consumers at these Kroger stores had boycotted the bags, Kroger, in addition to using boxes, could have put the unbagged groceries directly from the market basket into the shopper’s car on the first trip. On the second trip, the shoppers might have brought their own containers. Or Kroger *642might have found a way to increase its supply of boxes or it might have come up with some other solution such as furnishing market baskets temporarily or the big plastic trash bags that it sells. The point is that here the question, considered in context, is one of inconvenience, not impossibility, not financial disaster, not even serious economic harm. If consumers had complied with the boycott, Kroger could have, and no doubt would have, taken steps to ameliorate the problem.
Where an exercise of free speech is in question, I do not think we should be so quick to restrain it. In order for the expression to constitute a wrong against the neutral enterprise, the harmfulness of the words and conduct, as in “fighting words,” fraud and libel cases, should be clearer and the foreseeable injury more serious and direct. Although I have no problem with enforcement of the Board’s order without remand, a due respect for the First Amendment values should counsel us at the very least, if we are in doubt, to send the case back to the NLRB for reconsideration of the harm in light of Safeco.
. On the question of whether we must assume the “success” of the boycott, Judge Kennedy’s opinion for the Court in its rejoinder to my argument, contained in the last footnote of Judge Kennedy’s opinion, simply says that the Supreme Court in Safeco conclusively decided this issue, that Safeco is “dispositive.” That argument is wrong. There is no language in Safeco that addresses this issue directly one way or the other, certainly no language that could be said to be “dispositive.” The closest the Supreme Court comes to a discussion of the issue is in footnote 11 when it says the test is whether substantial loss is “reasonably likely.” It also says in the last line of the opinion that the First Amendment does not invalidate a law which prohibits picketing that “predictably encourages” (emphasis added) the boycott of a secondary enterprise. Certainly, neither phrase — “reasonably likely” and “predictably encourage” — supports Judge Kennedy’s view. Although neither phrase answers the question conclusively in favor of my position, both phrases suggest to me that we should assess the actual — the “likely,” the “predictable”— risk of harm, not just the abstract risk assuming success. This is what the NLRB did, and it seems to me that proper First Amendment analysis requires it. The Supreme Court has said nothing in Safeco to the contrary.