More on Unfalsifiable Claims – A Halakhic Perspective (Column 404)

In the previous column I discussed Popper’s criterion for the scientific status of a theory. We saw that he proposed a criterion according to which a theory is scientific if and only if it is empirically falsifiable. I noted that this statement has two faces: the objective—regarding the claim itself (a claim that cannot be falsified is weak), and the subjective—regarding the claimant (one who does not allow his claim to be subjected to a test of falsification is suspected of dishonesty). In this column I will present further facets and explain each of them. I will focus on various halakhic contexts and on their significance with respect to the falsification criterion.

Introduction to Logical Positivism

In the comments the positivist approach was mentioned—one that identifies the content of a claim with its empirical predictions. According to several logical positivists, a claim that has no predictions has no content. It is a collection of words that says nothing at all. Thus, for example, Carnap wrote this about all metaphysical claims (and also about ethics). For the logical positivists, Popper’s criterion is not a criterion for the scientific status of a theory but for its having meaning in the first place. In their view, what is non-scientific is meaningless.

This is a very extreme and unreasonable approach. The claim that there is a God is a claim with sense and meaning. The fact that it cannot be examined empirically does not render it meaningless. At most, if you are an empiricist, you can say there is no point in dealing with it because we cannot decide about it. But how can one say that it lacks meaning?! According to this view, even the claim that there are three billion ants in the world is meaningless, because there is no practical way to run an experiment that would confirm or refute it. One can of course argue that a hypothetical way to falsify the claim suffices, even if it cannot be implemented in practice—then the claim about the ants gains meaning. But to the same extent one can say there is a hypothetical and inaccessible way to verify that God exists or that fairies exist: simply equip ourselves with different senses/cognitive tools and encounter them.

Popper’s criterion is certainly reasonable as a basic requirement for determining the scientific status of a theory, but not for determining its meaningfulness. Even a theory that has no predictions can still have meaning.

Absence of Predictions that Stems from Lack of Meaning

Still, this conflation arises for a reason. Sometimes when Reuven advances a claim and Shimon says it is meaningless because it has no practical ramifications (nafka mina), he means that the absence of such ramifications is an indicator of its meaninglessness. The way to discover its lack of meaning is to look for a nafka mina for it. In other words, there are theories for which the absence of predictions truly results from their lack of meaning. They simply say nothing, and therefore have no predictions. My contention is that it’s clear that a meaningless theory will have no predictions (this is a prediction about meaningless theories), but a theory without predictions is not necessarily meaningless (and here the logical positivists err).

For example, someone proposes a theory that the people of Israel are a “chosen people,” i.e., unique and different from all other nations. The Gemara, and following it the Tur and the Beit Yosef at the beginning of Even Ha-Ezer (siman 2), even offered predictions for this: Jews are compassionate; from here they derived a rule that if someone is cruel we suspect he is not Jewish (there are stronger formulations: it is known that he is not Jewish). It seems to me that this criterion does not really stand the test of the facts. There are quite a few cruel Jews,[1] and I do not even think that our percentage of cruel people is the lowest among all nations. It is certainly not a decisive criterion. But beyond the truth of this claim, it may be that the claim itself does not actually say anything well-defined. What does it mean that a nation is different from all other nations? Every nation has statistical characteristics that differ from other nations. Does this claim say anything at all? I am not at all sure.[2]

The claim that this is a “chosen people” is not falsifiable, but I suspect that in this case the absence of a nafka mina stems from the claim’s lack of meaning. This is one example of what I argued above.

Another example is the well-known halakhic inquiry whether pikuach nefesh (saving life) permits Shabbat or merely overrides it (whether Shabbat is ‘hutrah’ or ‘dehuyah’ in the face of pikuach nefesh). As I understand it, this inquiry has no practical ramification. Usually two nafka-minas are suggested: that when one transgresses a prohibition to save life one should minimize the prohibition (“start with the lighter one”), or that one may transgress prohibitions not directly required for saving the life. Both types are brought by Rishonim and Aharonim, but when we examine the two definitions conceptually it seems there is no real distinction between the two sides of the inquiry. Even if Shabbat is permitted (hutrah) in the face of pikuach nefesh, it is still reasonable to say that only what is necessary is permitted. If a prohibition is unnecessary, or can be performed in a lighter way, there is no reason to allow doing it in the stricter manner.

I heard someone argue that the nafka mina is whether one is obligated to repent after having desecrated Shabbat for pikuach nefesh. If it is ‘dehuyah’, then ostensibly one transgressed and must repent. But this is, of course, nonsense. Not only did I do something permitted—I fulfilled my halakhic duty. So what should I repent for? Perhaps the Holy One, blessed be He, ought to repent for putting me in a situation where I was forced to desecrate Shabbat. Moreover, one component of repentance is accepting upon oneself never to return to that sin. Is that relevant here? Am I supposed to accept that next time I will leave a person to die rather than desecrate Shabbat? Absurd.

In my view there is no nafka mina between these two sides. Here too I suspect that the absence of a nafka mina stems from the claim’s lack of meaning. There is no real difference between the concepts of ‘hutrah’ and ‘dehuyah’, at least in this context. If I am right, then the absence of a nafka mina here stems from lack of meaning (or lack of difference in meaning between the two opposing claims). In a case like this, when I say that the claim that Shabbat is ‘hutrah’ has no nafka mina, I mean that it has no meaning. It asserts nothing beyond the claim that Shabbat is ‘dehuyah’.

Two marginal notes. One could perhaps argue with my specific claims here about ‘hutrah’ and ‘dehuyah’, but for my purposes it suffices to illustrate the general point: sometimes there can be claims that on the face of it seem meaningful, but further examination shows they are not. Beyond that, the absence of a nafka mina sometimes truly stems from lack of meaning (though not always—that is where the positivists err).

On the margins I will add that I have a sense that in the case of ‘hutrah’ and ‘dehuyah’ there actually is a distinct and different meaning to the two claims. The concept ‘hutrah’ is not synonymous with ‘dehuyah’. What I claim is that its application to the case of pikuach nefesh and Shabbat lacks real content. There is room to discuss whether in such a case there is no difference in meaning or whether this too is merely a lack of nafka mina.

Three Aspects of the Falsification Criterion

Thus far we have learned that the demand that a claim be capable of being subjected to a test of falsification has three facets:

• The meaning facet—sometimes a claim has no meaning if it has no predictions.
• The objective facet—a claim that supplies no predictions is weak in itself.
• The subjective facet—one who advances a claim that supplies no predictions is suspected of dishonesty, or at least of frivolity in making claims hastily and irresponsibly.

The discussion until now dealt with the meaning facet. A learned (lamdanic) claim is not suspected of falsehood, but the person who advances it is indeed suspect of making claims lightly and without responsibility. Primarily, this is a weakness of the claim itself and not of the claimant. When the two sides of an inquiry are identical, there is no lie or false claim here—rather an inquiry devoid of meaning. In Carnap’s terminology, this is a pseudo-claim (not in his original sense). But we saw that, contrary to the position of the logical positivists, the absence of prediction does not necessarily weaken the claim or reduce the value of dealing with it, and certainly does not empty it of meaning.

Back to Logical Positivism: The Meaning of a Nafka Mina[3]

The discussion of the example of ‘hutrah’ and ‘dehuyah’ was based on the search for a nafka mina to a learned-halakhic inquiry. The Gemara itself asks in several places “lemai nafka mina”—“what practical difference does it make?”—and seemingly there is an assumption that a discussion has no value if it has no nafka mina. Sometimes something is dismissed as “halakhta le-Meshicha,” i.e., relevant only in the Messianic era, meaning that it has no nafka mina now. Here it is clear that we are dealing with a meaningful claim (for it can have a nafka mina at another time). But even in cases where the Gemara wonders whether there is any nafka mina at all, we need not understand it as claiming that the discussion is meaningless. It may be that it simply lacks importance. Alternatively, the Gemara seeks a nafka mina in order to sharpen the discussion, but this does not necessarily assume that without a nafka mina the discussion is meaningless—and perhaps not even that it lacks importance.

We learn from here that the need for a nafka mina can be interpreted in two ways, just like Popper’s falsification criterion, each in its own context: (1) as a test of falsification; (2) as a test of meaning (in this context, of course, there is no question about the claimant’s honesty). When I claim that the Rambam took side A in some inquiry, I can subject my claim to a falsification test: I must check what the Rambam rules in a case that constitutes a nafka mina for the learned theory (that is its prediction), and that ruling will confirm or refute the theory. But there are situations in which I discuss a general inquiry, not necessarily in someone’s specific view. In such cases the existence of a nafka mina cannot serve as a test of falsification. If I did not find one, or I found an opposing view, the conclusion may be that a given authority did not hold so, but this does not rule out the very possibility of thinking that way. In such a case, the existence of a nafka mina is relevant to the discussion of the meanings of the two sides of the inquiry—namely, to clarify and sharpen that there really are two different sides and to elucidate the difference between them.

I will not expand here on nafka minas and their importance (see in my book cited below), and will suffice with one further remark on this matter. In yeshivot, it is common to examine the two sides of an inquiry by asking whether there is a nafka mina, or to dismiss it with the positivist claim that there is none. The yeshiva joke says that if someone presents two sides to an inquiry and his chavruta rejects it with “lemai nafka mina?”, the first can (and perhaps should) answer: “nafka mina for the betrothal (kiddushin) of a woman”—meaning, if a man betroths a woman on condition that side A of the inquiry is correct, then the question becomes whether she is betrothed or not.

Such a claim, of course, turns every discussion into a halakhic one and empties the demand for a nafka mina of content. The point seems to be that a nafka mina is not required in order to discuss. An inquiry is not merely a tool for clarifying laws in various cases; it has independent value. But that is not the whole picture. There is an important point here that is easy to miss. If there truly is no difference in meaning between the two sides (and it is not only a question of nafka mina), then there is no nafka mina for kiddushin either. One can offer the answer of kiddushin only when there is a clear difference in meaning between the two sides, but there is no practical ramification. By contrast, when there is no difference in meaning between them, then there will be no nafka mina for kiddushin either. In other words, the question “lemai nafka mina?” is positivist (since it identifies the meaningfulness of the discussion with the existence of a nafka mina), but the retort (nafka mina for kiddushin) is anti-positivist: it claims that the importance of a discussion is not contingent on the existence of a nafka mina. The real meaning of that retort is that if there is a difference in meaning between the two sides of the inquiry, then there is a point in conducting the discussion—even if it has no nafka mina.

A nice example of this appears in the Ran, Sanhedrin 15b. The Gemara there discusses how many judges tried the ox that approached the mountain at the Revelation at Sinai (the Torah says its sentence is death: “whether beast or man, he shall not live”): “The Sinai-ox—with how many [judges]?” In conclusion, from a verse in Ezekiel we learn that here too the rule applies that as the death of the owner so the death of the ox, i.e., it is judged to death by twenty-three judges.

The Ran there, in his novellae, asks:

And if you say: Why do we need this—whatever was, was? One can say: expound and receive reward. Alternatively, there is a nafka mina for someone who vows: if one says, “I am a nazirite if the Sinai-ox required twenty-three [judges],” if it did require twenty-three he is a nazirite, and if not he is not a nazirite.

He wonders, what practical difference is there, and answers two answers: (1) “Expound and receive reward” (i.e., it has the status of Torah even if it has no nafka mina). (2) A nafka mina for a nazirite (this is presumably the source of the yeshiva joke above).

I am not sure the Ran wrote this with irony, but it is certainly possible. In any case, whether written seriously or ironically, the meaning is the same: one does not need a nafka mina in order to discuss something (which, of course, brings us back to his first answer). But that is only if there is a nafka mina for kiddushin (or nazirite-vows), namely, if there truly are two distinct meanings to the two sides of the inquiry. Otherwise, the discussion has no value; it is empty semantics. What determines whether there are two different meanings is whether there is a nafka mina for a nazirite or for kiddushin. Of course, in this case the inquiry has two distinct sides with different meanings. What is lacking is only a nafka mina for our times (there was a nafka mina in the past; as we saw in the case of “halakhta le-Meshicha” it may be for the future).

Accordingly, the statement that an inquiry has a “nafka mina for kiddushin” is not a mere joke. It is a serious claim, saying that the lack of nafka mina in this case does not stem from the claim’s lack of meaning. The claim has meaning, and therefore it is worth discussing it, even if it has no nafka mina (at least for our times). This denies logical positivism, which identifies the meaning of a claim with the existence of predictions (nafka minas). A positivist could not tell the joke about a nafka mina for kiddushin. According to him, if there is another nafka mina, then the kiddushin one is unnecessary; and if there is no other nafka mina, then the inquiry has no meaning and thus there is also no nafka mina for kiddushin.

Explaining the Objective Aspect

On the face of it, the subjective aspect of the falsification criterion is understandable: “one who wishes to lie, distances his testimony.” But the objective aspect is not clear. Why does the fact that a claim cannot be falsified make it weaker? Suppose the one who advances it does so innocently (not lying)—does the fact that it cannot be falsified, in and of itself, weaken it? Why? Seemingly the claim remains in an even (undecided) state (either it is true or it is not, and we have no indication either way). Yet it is hard to deny the intuition that there is indeed some weakness here in the claim itself. This is what I called above the objective aspect.

To explain this intuition, let us look at the totality of claims in the world. We can divide them into true and false claims. I wish to argue that the number of true claims is far smaller than the number of false ones. This is a tricky point, since one could contend (!) that for every true claim one can define a claim that denies it, which of course will be false—and then it follows that the numbers are equal. But that is formal pilpul. I am speaking about positive claims (think of the form of a scientific law). When I say that the number of ants in the world is X, only one number substituted for X yields a true claim. Every other number yields infinitely many (countably infinite) false claims. If I say the gravitational constant is G, only one value substituted for it yields a true claim. Every other value (that is already an uncountable infinity) yields a false claim. And so forth. For every claim of the form of a scientific law there is only one true instantiation; all other instantiations are false.

If so, when I choose a claim at random, then before we have performed any experiment, the probability that the claim is true is negligible (that probability is the ratio of the number of true claims to the total number of claims). If I have performed several experiments that corroborate it, that increasingly strengthens the probability that it is true (this is the confirmation thesis from the previous column). Of course, there may still be accidental fit and the claim be false.

What shall we say now about a claim that is not falsifiable? If it is not grounded in some reliable source that knows the truth (the Almighty, a prophet, etc.), the probability that it is true is negligible. Remember: such a claim is by definition not subject to confirmation or to falsification. The correct attitude toward it is that, by default, it is a false claim (we saw that the probability that an unconfirmed claim is true is negligible). The status of such a claim is like that of a falsifiable claim that has not yet been tested. If there is a difference between these two kinds, it concerns only the claimant and not the claim (the subjective aspect). One who advances a non-falsifiable claim is suspect. One who advances a falsifiable claim is not suspect even before we have tested it. But in terms of the force of the claims themselves (the objective aspect), their status is similar.

As I hinted, there are claims that are not falsifiable but do have a reliable source. There is, of course, no reason to reject such claims merely because they cannot be subjected to an empirical falsification test. A claim about the existence of God, the Exodus from Egypt, and perhaps even the coming of the Messiah, are indeed not scientific and are not falsifiable, but it is not correct to reject them for that reason. The reason is that there are good a-priori grounds for adopting them, even without empirical confirmation. The existence of God is plausible in light of philosophical arguments, or due to a tradition about revelation, etc. The claim about the coming of the Messiah was conveyed to us by the Almighty, who certainly knows this. The Exodus is testimony transmitted to us by reliable witnesses. Without entering into debates about each of these examples, they suffice for me to illustrate the principle that non-falsifiable claims are not disqualified by that fact alone.

As a marginal note I will add that if we are dealing with a specific claim, like the coming of the Messiah, then ostensibly there are only two possibilities: that he will come or that he will not. But even in such a case, if I have no specific information from a reliable source, the presumption is that he will not come. That is, even though in such a case there is no difference between the number of true and false claims, and ostensibly the probabilities of truth and falsehood are equal, still in such a situation the burden of proof lies on the one who claims that the Messiah will come (this is exactly like Russell’s celestial teapot). Of course, this is all the more so regarding claims that are implausible on their face. Often implausible claims are also non-falsifiable (otherwise they would already have been falsified), and therefore there is a tendency to reject them for that reason. But that reason is unnecessary. There is simply no reason to accept such claims—not only because they are non-falsifiable, but mainly because they are implausible.

Interim Summary

When I examine a claim presented to me, I should ask several questions, in order: (1) Does it have meaning at all (or is it a pseudo-claim)? If yes, then: (2) What is its initial source (are there good a-priori grounds to adopt or reject it, or is it hanging in the air)? If not, then: (3) Is the one who advanced it suspect (is it falsifiable)? If yes (and perhaps even if not), then: (4) Has it been corroborated or not, and to what extent.

From this summary we can see that falsifiability is a significant criterion, but it is far from being exclusive, or even decisive. It is also important to note the three aspects of falsifiability (meaning, objective, and subjective), which do not always appear together. In any case we must discuss the aspects relevant to it. Now let us look at several halakhic examples.

“Bari” and “Shema”: When It’s Worse

In one of the comments to the previous column the example of “bad bari and shema” arose. The Ba’alei ha-Tosafot in several places distinguish between two types of “shema” (maybe) claims. Think of a situation where someone sues me for a loan I took from him, and I respond that I do not remember (a shema claim). In such a case I ought to know whether I took a loan, and if I answer weakly, this arouses suspicion that I may be trying to evade payment with a gentle lie. Therefore this is a “bad shema.” But if a depositor claims as a “shema” that the animal he deposited with a paid custodian was lost (a paid custodian is liable for loss), this is a “good shema.” The depositor cannot know what happened to the animal, since he is not at the custodian’s home (this is apparently why the Torah requires the custodian to swear if he denies the depositor’s claim).

There is a similar distinction between two types of “bari” (certain) claims. When the custodian claims with certainty that the animal was taken by force (ones) and is therefore exempt, this is a “bad bari.” The opposing side does not know the truth and cannot refute him, so it is easy for the custodian to assert confidently that that is what happened—even if he is lying. By contrast, when a lender sues the borrower with a certain (bari) claim, the defendant is expected to know the truth and can refute him and expose him as a liar. When the lender makes a bari claim in such a situation, he takes the risk of being contradicted, and therefore this is a “good bari.”

This distinction indeed reflects the fact that one who advances a claim that can be refuted takes a risk, and that strengthens his standing. This is not support for the claim itself but for the claimant’s credibility. Here the focus is subjective, not objective.[4]

Testimony That Cannot Be Disproved (Hazama)

Another example of a similar consideration is the halakhic rule that any testimony that cannot be subjected to hazama (disproval by “you were with us elsewhere”) is inadmissible. The similarity between this halakhic rule and Popper’s falsification principle is already noted in my Quartet. It was treated extensively by the couple Yemima and Hanina Ben-Menachem (she a philosopher, he a legal scholar) in their article on “testimony that cannot be subjected to hazama” and the falsification principle. The points they discuss there, however, are not relevant to our discussion here.

As for the rules of evidence in halakha, when we have two witnesses this is the strongest possible evidence, and nothing—not even a hundred witnesses—can topple the testimony of two. Therefore, when there are two witnesses pitted against a hundred, in halakha this is considered “contradicted testimony” (we are left with a tie, i.e., doubt). The only way to topple testimony[5]—that is, to prove that the witnesses are liars—is by way of hazama: when two other witnesses come and say that the first witnesses were with them in another place at the time and therefore could not have seen what they testified about, the court accepts the testimony of the latter (the “hazamim”) and the first are declared liars and are even punished for false testimony.[6]

Beyond the definition of the hazama procedure itself, there is also a halakhic rule that any testimony must be capable of being subjected to hazama. Testimony that cannot be subjected to hazama is inadmissible (see Bava Kamma 75b and elsewhere). There are different situations in which it is impossible to subject testimony to hazama. For example, if the witnesses did not give us the exact time of the event, one cannot say to them that at that time they were elsewhere and could not have seen it (this strongly resembles the Delphic oracle and the vague predictions discussed last column). Alternatively, when for some reason it is impossible to punish the witnesses even if they are subjected to hazama, this too counts as testimony that cannot be subjected to hazama.

Seemingly, the hazama procedure is the basic mechanism, and based on it halakha rules that testimony that cannot be subjected to hazama is inadmissible. But it seems natural to conjecture that the situation is the reverse: the definition of the hazama procedure is intended in advance to increase the credibility of witnesses. If there were no way to subject testimony to hazama, then testimony that in fact was not subjected to hazama would not be so strong (it was not subjected to hazama because it could not be). To increase the credibility of witnesses, the Torah had to define a procedure by which their statements could be rejected, and therefore it defined the hazama procedure.[7] The fundamental principle is that testimony that cannot be subjected to hazama is inadmissible. The hazama procedure itself—namely, granting precedence to the hazamim over the hazumim—is a secondary principle that serves the fundamental rule.

Either way, it appears that the very existence of a hazama procedure turns testimony into a claim that can be refuted, thereby strengthening it. This brings us very close to Popper’s falsification criterion. There too we saw that the possibility of refutation strengthens the claim (the objective dimension) and also strengthens the standing of the claimant (the subjective dimension). In the halakhic context, if the testimony can be subjected to hazama it is stronger, since the fact that it was not subjected to hazama corroborates it. If it cannot be subjected to hazama, then the fact that it was not is trivial. Beyond that, when testimony can be subjected to hazama, the witnesses’ standing is also stronger, because they are not suspected of advancing claims that cannot be refuted (they are taking a risk). I note that the first example I gave above—testimony that does not provide a precise time—reflects a weakness in the content of the testimony itself (its content is thin and ill-defined). The second example—when it is impossible to punish the witnesses—reflects a weakness of the witnesses (they are not taking the risk of being punished).

Needless to say, the fact that we fear such witnesses may be lying does not, of course, mean that such testimony is false (we will not declare witnesses who cannot be subjected to hazama liars, nor even suspect them of lying). The point is that the lack of possibility to subject them to hazama creates a situation that enables lying more easily. Put differently, the barrier to lying in such a situation is not strong enough, and therefore such testimony does not have the absolute standing that regular testimony has.

Clearly, even staunch positivists would not claim that such testimony is contentless or meaningless, for the impossibility of hazama is unrelated to the meaning of the claim itself. A claim that so-and-so murdered someone in some month (and not on a specific day), or a claim that consequently one cannot punish the witness, are claims with clear meaning—even though they are not falsifiable.[8] Therefore, in the example of testimony that cannot be subjected to hazama, the subjective and objective aspects appear, but not the meaning aspect.

A Short Halakhic Discussion

Still, it must be said that the similarity between this halakhic rule and the falsification criterion is far from self-evident. There are situations in which the application of this rule seems entirely formal and unrelated to the question of falsification. I will bring one example.

There is a halakhic rule that a witness does not become a judge. One explanation for this rule appears in Tosafot, s.v. “kegon,” Bava Kamma 90b:

And the reason that one who testified cannot become a judge—some explain that it becomes testimony that cannot be subjected to hazama, since as he himself is witness and judge he will not accept hazama against himself.

Simply put, according to this explanation a witness does not become a judge because his testimony is not subject to hazama (he will not accept hazama against himself).[9] But note that at the time of his testimony he did not necessarily know that he would become a judge; therefore, at the time of testimony, his testimony was indeed subject to hazama. In terms of the requirement that the witness take a risk upon himself—he certainly did so, and only after the testimony did a situation arise in which he cannot be subjected to hazama. If the requirement that testimony be subject to hazama parallels Popper’s falsification criterion, here it would seem we have testimony that entailed risk and therefore should have been admissible.

Admittedly, this reason for the rule that a witness does not become a judge is not agreed to by all Rishonim, and therefore one may perhaps preserve the comparison to Popper’s criterion according to the other approaches. In general, it has been discussed whether, when a witness becomes a judge, what is disqualified is the testimony or the judgment (the witness or the judge). If we are dealing with Popper’s falsification criterion, then it is clear that the testimony is what is disqualified in such a situation.

There are additional examples that cast doubt on this comparison, and this is not the place to expand on them.

[1] See an example of this in the Shulchan Arukh itself, Even Ha-Ezer §71:1.

[2] See at length on this in the second book of my trilogy, Ein Adam Shalit Ba-Ruach, the sixth conversation in chapter 22.

[3] See a long and detailed discussion of this in the eleventh book in the Talmudic Logic series: The Platonic Character of the Talmud.

[4] In a pamphlet of mine on migo I explained that the ‘migo’ proof also concerns the claimant and not the claim, and that is its weakness (therefore, according to most views, we do not say “migo” to extract money). The fact that I claim X and could have claimed a better claim Y shows that I am not lying. But that is evidence of my credibility as a claimant, not of the truth of the claim itself. Thus, for example, if there is a concern of “bededemi,” namely that I am not lying but imagining, then the fact that I have a migo is irrelevant. The migo proves that I am not lying, but the concern that I am imagining remains. By contrast, presumptions (chazakot) are proofs about the claim itself and not about the claimant. For example, the presumption that a person does not repay a loan before its due date shows that the borrower’s claim that he repaid early is dubious. Such evidence concerns the claim and not the claimant.

[5] I will not enter here into “the slain man walks in” or admission of a litigant. These are two situations in which even the testimony of two witnesses can fall.

[6] For the reasons for distinguishing between hazama and contradiction, see my article on “scriptural decree.”

[7] This does not mean that the preference for the second set of witnesses over the first in hazama is arbitrary. But it may be that the preference would not suffice to always accept the second set and declare the first liars, and we do so in order to subject the testimony to a hazama test.

I note that the Ben-Menachem couple, in their article cited above, compare Popper’s falsification principle to the halakhic rule disqualifying testimony that cannot be subjected to hazama. They present there a different reason for defining the hazama procedure. In their view, it comes to resolve the concern about a witness who testified falsely in innocence (mistakenly). They implicitly assume there that even in contradiction (hakchasha) the second testimony prevails, and the reason we do not accept it is that the first may have been given innocently. In hazama, by contrast, it is clear that the false testimony was given willfully (since they were not at the scene at all). This is by no means necessary (even in hazama they may have innocently erred about the date), and it also does not fit the Talmudic sources. From the sugyot it clearly emerges that in contradiction the two sets have equal standing (and not that the second is preferred), and it is quite clear that this has nothing to do with whether the lie was innocent or willful.

[8] Theoretically, one could find witnesses whose hands never left those of the witnesses to be hazama-ed for the entire month, and they could subject them to hazama. But this is only a hypothetical possibility.

[9] In Tosafot’s wording it is not explicitly stated that the basis of the disqualification is that the testimony cannot be subjected to hazama. One could have explained that we disqualify the judge because he will not accept hazama against the testimony, and not that we disqualify the testimony because it cannot be subjected to hazama.

But from the continuation of Tosafot it emerges that the reason to disqualify a witness who becomes a judge is indeed because of testimony that cannot be subjected to hazama:

And this reason is difficult for the Ri of Orléans, for if so, witnesses who are relatives of the judges should not testify before them, since it would become testimony that cannot be subjected to hazama. Rather, necessarily they can be subjected to hazama in another court; so too here, they are capable of hazama in another court.

Tosafot do indeed reject this reason, but it is clear from their words that the meaning of the reason is that we are dealing with testimony that cannot be subjected to hazama.