The Halakhic Status of Hermeneutical Methods – Analogy and Induction in Jewish Law

Tzohar – 5763

Induction and Analogy in Jewish Law: An Analysis of Maimonides' Second Root

A. Introduction

In a previous article (Tzohar 12; hereafter: Article A), the logical status of the methods of derash was discussed. In the second root, Maimonides disputes with Nachmanides regarding the halakhic status of the laws derived by means of these hermeneutical methods (Torah-level or rabbinic). In this article I will try to show the connection between the two planes: the logical and the halakhic. For the sake of what follows, I will briefly present the conclusions of Article A.

There we saw a dispute between Ralbag, who holds that the rules are vague and therefore can serve only as support for known laws (and not create new laws), and Maimonides, who holds that the methods of derash are univocal, and can therefore also be used to create new laws (and not merely support accepted laws). We noted that from Maimonides it appears that the rules themselves have no philosophical significance, and are merely a code, or cipher, arbitrary in its essence, for generating laws from Scripture.

By contrast, we saw there the approach of Rabbi HaNazir, according to which the rules themselves do have philosophical significance. The core of his innovation was that although the rules are analogical (or inductive), this does not mean that they are vague. A person (the Jew) possesses an ability—what he calls 'auditory logic'—to grasp analogies sharply and to distinguish correct analogies from incorrect ones.

We also noted there that there is a difficulty regarding the relation between analogy and induction. An analogical inference, for example, is learning from the fact that frog A is green that frog B is also green. A parallel example of inductive inference (from particulars to the general rule) is generalizing from the fact that we observed several green frogs to the general law that all frogs are like that.

We noted that induction is ostensibly based on many analogies and gathers them into a general law. After learning of each frog separately that it is green, we generalize and determine that all of them are so. If so, analogy would seem to be the fundamental form of inference, and induction merely a collection of analogies. On the other hand, every particular analogy between frog A and frog B seemingly presupposes an induction, since their similarity rests only on the fact that B too is a frog. That is, we are not learning about frog B as such, but about frogs in general, and in particular frog B. From this the opposite conclusion follows: precisely induction is the basic inferential process, and analogy is only its application to a particular case.[1]

In this article I wish to examine the views of the medieval authorities (Rishonim) regarding the nature of midrashic analogy, and the relation between analogy and induction. Through this topic we will also see the connection between the logical status of the methods of derash and their halakhic status. We will also try to show several central implications of these approaches that touch on the foundations of Jewish law, such as the understanding of the relation between the plain meaning and the interpretive exposition, the relation between Torah-level and rabbinic laws, and more.

B. The Dispute between Maimonides and Nachmanides in the Second Root

Maimonides opens the root with a distinction, also presented in his introduction to the Commentary on the Mishnah, between 'creative' and 'confirmatory' derivations (in Article A: supportive derivations). Maimonides clearly presents the position that among the midrashim there are also creative midrashim. In Article A we saw that this is Maimonides' position elsewhere as well (as noted, Ralbag represents the opposite view).

At the next stage of the discussion, Maimonides presents his main innovation: the laws yielded by 'creative' midrashim are 'Divrei Sofrim' (rabbinic in force), whereas the laws arising from 'confirmatory' midrashim are Torah-level.[2] Maimonides' basic assumption is that a midrash is creative unless the Sages who expound it explicitly say that it is confirmatory. The way they say this is by determining that the law derived from it is Torah-level. Maimonides explains that in such a case we conclude that the midrash confirms a known law transmitted by tradition from Sinai (and does not create it), and therefore that law has Torah-level force.

This position of Maimonides raises several difficulties:

A. Maimonides' determination that there are two kinds of derash. In Article A we raised a difficulty against Ralbag: what purpose is served by a confirmatory midrash if the law was received by tradition? Here one may ask the reverse as well: what purpose is served by transmitting laws orally from Sinai if they can be derived from the verses in a clear and authoritative way (for Maimonides, as noted, the methods of derash are not vague).

B. The determination that the laws derived from scriptural midrashim have the status of rabbinic laws. If these methods are indeed a reliable and authoritative interpretation of Scripture, how can what emerges from them be classified as rabbinic law? These laws would seemingly be latent in the verses, and part of the Torah's intent. This question is raised by Nachmanides in his glosses, and Maimonides himself also addresses it; see below.

C. Another difficulty raised by Nachmanides is that Maimonides holds that a law given to Moses at Sinai also has merely rabbinic force.[3] If so, asks Nachmanides, how can one maintain that a law transmitted by Moses at Sinai is rabbinic, and a law derived through the thirteen hermeneutic rules is rabbinic, but the combination—namely, a law derived through the thirteen rules and also transmitted from Sinai—is Torah-level.

Yet in light of difficulty C, it seems possible to understand the answer to A. According to Maimonides, the reason a law is transmitted to Moses at Sinai even in a case where it can also be derived from scriptural midrashim, and vice versa, is in order to indicate that the status of this law is Torah-level rather than rabbinic. As we have seen, in his view each of these mechanisms on its own creates rabbinic law, and only their combination can create a Torah-level law.[4]

At the outset one should note a surprising conclusion. According to Maimonides' conception, the notion of 'Torah-level' has a different interpretation from the accepted one (including among Maimonides' commentators and scholars). Usually it is explained that Maimonides classifies as Torah-level all laws conveyed to Moses by the Holy One, blessed be He, at Sinai (see, for example, Halbertal), and that this classification expresses the difference between Moses' prophecy and the prophecy of all other prophets. Our discussion here suggests that Maimonides' conception is altogether different: a 'Torah-level law' is a law explicitly found in the Torah. Laws conveyed to Moses at Sinai directly by divine revelation, such as the laws Maimonides calls 'a law given to Moses at Sinai,' are not necessarily Torah-level laws. According to Maimonides, only what is explicitly found in the Torah is a Torah-level law.[5]

This is explicit in Maimonides' aforementioned responsum, where he writes:

And nothing there is of Torah status except something explicitly stated in the Torah, such as wool-linen mixtures, forbidden mixtures, the Sabbath, and forbidden sexual relations…

The first difficulty above is technical in essence, and can therefore be understood as we have seen. Yet two substantive difficulties still remain in Maimonides' view (B and C above). In light of what we have said here—namely, that in Maimonides' doctrine the notion of 'Torah-level' is a category describing laws explicitly stated in Scripture—we need a more precise definition of the phrase 'explicit in Scripture,' and through that perhaps we will understand the classification of midrashic laws as rabbinic (B above). In what follows we will examine this topic, and from the discussion it will also become clear how the combination of the two criteria creates, as it were out of nothing, Torah-level laws (difficulty C).

Further on, Maimonides attacks Bahag, who represents those who hold that the laws derived from halakhic midrashim have Torah-level force, in two main directions:

A. A verse does not depart from its plain sense. The Talmud asks in several places, after a halakhic derash is cited, what the plain meaning of the verse is, for a verse does not depart from its plain sense (see, for example, Babylonian Talmud, Yevamot 11b). Maimonides asks: if the midrash truly exposes the meaning of the verse, as Bahag holds, why does the Talmud not suffice with the midrashic interpretation of the verse?

According to Maimonides' assumption, there is only one adequate interpretation of a verse, and it is precisely on this point that Nachmanides disputes him in his glosses. Nachmanides maintains that different interpretations of the same verse are entirely possible, and all of them can be true.[6]

Another question raised by Nachmanides is the status of halakhic midrashim as interpretations of Scripture. If Maimonides maintains that these are true interpretations of the verse, why should they not be considered legitimate interpretations and counted as independent Torah-level commandments? And if, on the other hand, they are not true, then they have no place at all. The Sages can institute enactments and decrees, but halakhic midrashim are interpretations of Scripture and not ordinary rabbinic laws.

Maimonides himself already senses this question and rejects it by saying:

Perhaps you think that I refrain from counting them because they are not true. Whether the law derived by means of a rule is true or untrue—that is not the reason. Rather, the reason is that everything a person derives is branches that emerge from the roots that were stated to Moses at Sinai in explanation, and these are the 613 commandments. Even if the one deriving them were Moses himself, it would not be proper to count them.

It seems to me that in this passage lies the key to understanding Maimonides' novel position. In the next chapter I will try to examine these remarks and, on their basis, understand the points that remained problematic in Maimonides' discussion of the second root, as described in this chapter.

C. 'Branches Issuing from the Roots': On the Question of the Nature of Midrashic Analogy

In the introduction to the Commentary on the Mishnah, as in other places, Maimonides states that the thirteen hermeneutic rules were transmitted to Moses at Sinai. If so, Nachmanides' argument would seemingly be justified: how can one say that the use of interpretive methods transmitted at Sinai, applied to a text likewise given there, gives rise to laws whose force is only rabbinic? Nachmanides assumes that midrashic laws are found, albeit indirectly, within the text itself and constitute an authoritative interpretation of it. We saw that Maimonides too, in addressing this question, states that the laws derived from these rules 'are not untrue,' and thus the question becomes even stronger: why is their force only rabbinic?

Halbertal proposes a classification based on the nature of the outcomes of halakhic midrash. A midrash whose purpose is interpretive—that is, one that merely clarifies the meaning of expressions in the text—has Torah-level force, whereas a midrash that infers additional external conclusions from the verses has rabbinic force. To illustrate this distinction, Halbertal discusses an imaginary law stating that vehicles may not be parked in a public garden. Interpreting what counts as 'vehicles' for purposes of that law would be an interpretation of the law itself, and therefore would have the same status as the original law (= Torah-level). By contrast, a ruling that derives from that law a prohibition on bringing vehicles into schools as well would be a case of 'branches issuing from the roots.' That is not an interpretation of the law but an attempt to extend it into additional rules (branches) that are not contained within it, and therefore its legal conclusions do not have the same force as the original law (they are rabbinic).

Halbertal's distinction is based on the aims, or results, of the act of derash; in other words, it reflects the accepted distinction between explanatory midrash and creative midrash.[7] It seems to me that this distinction indeed well explains the expression 'branches issuing from the roots,' but with respect to the other problematic points described above it does not suffice. Beyond that, it says nothing about derash in its essence, but only distinguishes between different kinds of derash. Therefore Maimonides' distinction is not entirely clear, for his remarks refer to all midrashim as such. The explanation of why the creative midrash, by the very definition of its methodological character, yields rabbinic laws still requires further clarification. I wish to add here a further elucidation of Maimonides' method, in light of the definition we saw of the notion 'Torah-level' in his doctrine, a definition that makes the difference depend on the character of the method of derash itself, and not only on its aims.[8]

In Article A we discussed the analogical character of the methods of derash. We also saw there that deduction is characterized by the fact that its conclusion is in effect already hidden in its premises, and therefore it is only the exposure and analysis of what is already known to us, bringing it to light (or into actuality). Analogy and induction, by contrast, are not an uncovering of the known, but an inference of additional and new conclusions from known premises.

According to our proposal, Maimonides' approach is based on the fact that halakhic derash is an inference from Scripture that is carried out by way of analogy. A deductive analysis of Scripture exposes what is hidden within it, and in this sense it is only interpretation. The results of a deductive inference from a verse will of course have the force of Torah-level laws, since such an inference uncovers what was hidden within it, and according to Maimonides what is truly inside Scripture is 'Torah-level.' But the results of analogical inferences from Scripture cannot be regarded as literally present within it; rather, they are an extension of the verse by way of analogical resemblance. Conclusions inferred in this way are not considered by Maimonides to be literally present in the verse, and therefore they cannot be regarded as having Torah-level force.

Nachmanides holds that even the result of an analogical inference from the Written Torah counts as an interpretation that is contained within it, so long as the hermeneutic methods that led to it are legitimate—among those given at Sinai. Therefore, as we saw in the previous chapter, Nachmanides holds that there may be two legitimate interpretations of the same text of the Written Torah, whereas for Maimonides this is impossible. According to Maimonides, any interpretation that is not the plain meaning is not literally present in the text.

There is no doubt that Nachmanides also understands that the methods of derash are analogical, and on that point he does not disagree with Maimonides. It is possible that he defines the notion of 'Torah-level' differently. Yet Nachmanides disputes Maimonides even on Maimonides' own terms. He argues against him that if the rules were given at Sinai and the Torah was given there, then the combination of the two necessarily produces laws that are explicitly present in the text. We must therefore clarify more precisely the foundation of the dispute between him and Maimonides.

For this purpose we must pay attention to another assumption (beyond the analogical character of the methods of derash) that is embedded in Maimonides' position, and we will try to explain it by first introducing an important distinction in the philosophy of science (and in philosophy generally).

Kant distinguished between synthetic and analytic propositions. An analytic proposition is one that is derived from the very definition of its subject. For example: this ball is round. This proposition adds no new knowledge about the world beyond what is contained in the definition of the subject. To know it, it is enough for us to analyze the subject of the proposition. By contrast, a synthetic proposition is one that adds knowledge beyond the very definition of its subject. For example: this ball is heavy. Weight is not an essential property of balls, and therefore this adds new knowledge beyond what is contained in the definition of the concept 'ball' (which is the subject of the proposition).

The accepted approach in philosophy is that analytic propositions are a priori; that is, we do not need experience in order to recognize their truth. The philosopher Saul Kripke challenges this conception by saying that there are analytic propositions for whose truth we nevertheless need experience in order to become aware of them. For example, the proposition 2+2=4 is an analytic proposition. That fact does not emerge from experience, but from the very definition of the concepts involved in the proposition (the numbers 2 and 4). Nevertheless, in order to teach the child this analytic fact, the teacher uses empirical demonstrations (combining objects). It may even be that without the demonstrations it is impossible to learn this proposition at all.

This challenge teaches us that there are situations in which we learn an analytic fact—that is, one embedded in the very definition of its subject—but the way in which we learn it is by analogy, and not by analyzing the subject under discussion.

Let us take a scientific example. In Newtonian mechanics we learn, by empirical methods and through the use of analogies and inductions, that when a force acts on a body possessing mass, it accelerates with an acceleration proportional to the force. This is Newton's second law. Even after we have learned this law by means of analogy or induction, one can still think that it is analytically derived from the concept of 'mass.' It is possible that, had we been able to understand the concept of 'mass' in all its depth without experience (as in the previous example regarding the numbers 2 and 4), we could have inferred the second law a priori from that understanding. If so, the second law is derived analytically from the concepts of force and mass, except that owing to our limitations we learn it by analogies rather than by conceptual analysis alone.

One may imagine an even more general conception, according to which once we understand any law of nature, we always discover that it is really an analytic proposition. According to this philosophical conception, the laws of nature that are learned by various analogies and inductions may themselves be analytic deductions.

The conclusion is that there is no necessary connection between the way a person infers a certain conclusion from some assumption, and the real relation that exists between the assumption and the conclusion. Analogy describes the way the person learns, but not necessarily the relation that exists between the premise and the conclusion in themselves.[9]

Let us now return to the topic of the methods of derash. When we use analogical methods of inference in order to create laws from the verses, we have described only the way the expounder learns those laws. One can still say that the laws produced in these ways are indeed literally present in the verses that are being expounded.

To understand Maimonides' position in the way it was presented above, it is not enough to assume the simple and accepted fact that the methods of derash are analogical rather than deductive. We must add the further assumption that the result produced by the midrash is itself related to the plain meaning of Scripture only by way of analogy, and is not literally present in it. It is highly plausible that the determination that the law is rabbinic depends on the relation between it and what is written in the Torah, and not on the question of how we arrived at that law.

Nachmanides' position is that although the methods of halakhic midrash are analogical, the conclusions of the derash are contained in the text itself, and in principle could have been derived from it by conceptual analysis. Consequently, their force is Torah-level.

If so, the first difficulty raised above in Maimonides—why these laws are rabbinic—is understandable in light of his own criterion, according to which only laws literally present in the text have Torah-level force. The laws created by midrash are not literally present in the text.

We now turn to the second question: why, when both requirements are present (tradition and halakhic derash), is the law indeed Torah-level, whereas each of these requirements on its own is insufficient for that.

In light of the conception of Maimonides presented here, a law that emerges from scriptural derash is not perceived as being contained in the text, but rather as an extension of it, something quasi-analogical. The reason for this, as stated, is that Maimonides understands analogy not merely as the way in which the expounder arrived at his conclusion, but as a relation that exists between the laws themselves. The conclusion of the derash is not present in the verses themselves, but constitutes an extension of them.

D. Analogy and Induction in the Hermeneutic Rules

We saw in the introduction that there is a dilemma concerning the relative status of analogy and induction. Is analogy the more basic process, standing at the foundation of induction as well, or is it the other way around? One can formulate the dilemma regarding analogy as follows: is there an independent analogical process behind which no induction stands? Or: can one discern a similarity between two particulars without passing through the general category common to them both? If such an analogical process indeed exists, then analogy is not necessarily built upon a hidden induction. If no such process exists, then every analogy is based on a hidden induction. Parallel questions can of course also be asked regarding induction. There is, naturally, also a third possibility, according to which these two processes are independent of one another, and both are autonomous. In what follows we will find support for all three of these approaches among the commentators.

In the simple and prevalent conception of analogy, a general principle is always hidden behind it. When I see that Jacob is mortal, I do not see merely a certain creature who is mortal; rather, I perceive that human beings are mortal. This recognition is called, in the phenomenological doctrine of the philosopher Edmund Husserl, 'eidetic vision': through the particular, which is as it were transparent, I see the universal.[10]

Let us now apply this mode of analysis to biblical interpretation. When I study the command 'You shall fear the Lord your God,' I am in effect learning the general principle that obligates one to stand in awe of anyone essentially bound up with Torah, and from that I infer that one must also stand in awe of Torah scholars. From such a perspective, the command to fear Torah scholars is explicitly written in Scripture. The form of writing is by means of a particular case (fear of God Himself) that exemplifies the general principle behind it.

We saw that Nachmanides understands the command to stand in awe of Torah scholars as explicit in the text. He apparently holds that analogy is always made by way of grasping the general principle behind the particular example; otherwise analogies could not be made at all. Therefore, when the Torah says 'You shall fear the Lord your God,' this is merely a way of writing in particular form the general principle that commands awe of anyone connected to Torah. From this conception Nachmanides reaches the conclusion that awe of the sages is explicitly written in Scripture within the command to fear God, for the Torah is in effect commanding fear of anyone connected to Torah. The analogy made by the expounder merely serves as a tool by means of which he uncovers the general principle to which the Torah itself refers, that is, a tool for generalization, or induction.[11]

Maimonides, by contrast, apparently understands midrashic analogy as proceeding directly between particulars. There is no general rule behind it that includes both particulars (the source case and the derived case). When Scripture says 'You shall fear the Lord your God,' the meaning is to fear Him alone. The sage who expounds the verse senses in his intuition that Torah scholars, too, must be revered, because there is something about them that resembles their Creator. But this is not derived from any comprehensive principle found in the intent of the verse itself. It seems that Maimonides adopts here an expansive interpretation of the principle that we do not expound the reason of the verse, according to which one may not determine even what generalization, if any, underlies the particular command in the verse.

It follows that Maimonides and Nachmanides understand the mode of operation of midrashic analogy differently. For Nachmanides, analogy is based on a hidden induction, and therefore uncovers a general principle that is present in the text, whereas Maimonides maintains that precisely analogy is the more basic tool, and there is no hidden induction behind it.

We can now perhaps understand the final difficulty that remained in Maimonides' view (C above). When there is a tradition supporting the law that emerges from scriptural derash, it teaches me to regard the analogy I employ in the process of derash as based on induction. Without tradition I have no logical ability to generalize the particular example learned from the verse into a general law (for we do not expound the reason of the verse). Consequently, in such a case the learned law is not contained in the verse but constitutes an extension of it, and its status is that of a rabbinic law. If there is a tradition supporting the derash, that means that I may assume that at the basis of the midrashic analogy stands a general law, and consequently the conclusion of the derash is considered explicit in the text (as occurs, according to Nachmanides' view, in all such derivations).

One can find among writers on the methodological rules support for both of these approaches. It is customary to classify the midrashic process called 'mah matzinu' as analogy, whereas induction is the rule called 'binyan av.'[12] The term 'av' ('father'), which appears in the name of this rule, is usually interpreted as pointing to the teaching law, which is the 'father' of the law that is learned (the 'son').[13] Yet in light of the determination that 'binyan av' is an inductive midrashic process, one can also understand that the 'father' is the general command or principle that underlies the two particular laws (the source and the learned case: the two 'sons'). What lies beneath the source law is in fact a general law, and that is what is called the 'father' of the learned law.

The rule of 'mah matzinu' is not mentioned explicitly among Rabbi Ishmael's thirteen rules in the baraita of the thirteen rules at the beginning of Sifra. However, in Sefer Halikhot Olam, Gate 4 (the laws of binyan av), it is brought in the name of some that this is the rule called in Rabbi Ishmael's baraita 'binyan av from one verse' (or: 'and one verse'). This is also the opinion of Rabbi Shimshon of Kinon in Sefer HaKeritut, part I (Batei Middot), House 3, Chamber 2,[14] and see there for his proofs. However, the opinion of the Ra'avad (cited there), in his commentary to the above baraita, is that 'binyan av' applies both to one-from-one and to one-from-two. So too Rabbi Aharon ibn Hayyim understood in his book Middot Aharon, chapter 4 on the rule of binyan av, that Rabbi Shimshon and the Ra'avad disagreed on this point (and his own view inclines toward that of Rabbi Shimshon).[15]

To summarize, the Ra'avad's view is that even learning one case from one case is 'binyan av,' whereas for Rabbi Shimshon only one-from-two is called 'binyan av,' and one-from-one is 'mah matzinu.'

It seems that there is here a dispute precisely about our issue. When one learns one case from two, one in effect requires two source cases in order to extract the feature common to them, and in that way one can learn that the rule applying to them does not characterize only those cases, but all cases possessing that feature. This, as stated, is an inductive generalization. By contrast, when one learns one case from one, this can indeed be understood as an induction based on a single source case, but more simply it seems preferable to understand it as a direct analogy from one particular to another.

The need for two source cases in a 'binyan av' of one-from-two exists precisely because of the difficulty to which I pointed above. From a single particular it is not always possible to extract which feature is the one relevant to the inductive generalization. When one examines two examples and investigates what they share (the common denominator), one can identify with greater confidence the correct generalization.[16] Analogy from particular to particular, by contrast, is performed intuitively and does not presuppose a generalization (induction) at its foundation, and therefore it does not require two source cases.

It follows, then, that according to Rabbi Shimshon, analogy does not pass through a hidden induction but proceeds directly from one particular to another, and therefore one-from-one cannot be called 'binyan av.' We are not seeking a 'father' here; rather, we learn from one 'son' directly to another. By contrast, the Ra'avad apparently understands that every analogy is based in essence on a hidden generalization, and therefore even learning one-from-one (= analogy) is included within 'binyan av,' that is, this too is induction.

Induction, however, according to Rabbi Shimshon of Kinon, is carried out only on the basis of two source cases, because extracting their common feature is a necessary indication of the truth of the inductive generalization. Clearly, in the case of 'binyan av from two verses' we are dealing with two laws between which there is mutual necessity—that is, one cannot learn one from the other and vice versa. In such a case we make a 'what is common to them' argument, and only then is this the rule of 'binyan av from two verses.' If there is no mutual necessity between them, then the second law adds nothing at all to the first, and therefore it does not help us test our halakhic generalization by means of an additional case. Two laws between which there is no mutual necessity add no logical force to the inductive generalization beyond that provided by a single law. If such a law nevertheless appears (one that is unnecessary), then the two are in the category of 'two verses that do not teach' (or even teach the opposite, and the matter is well known).[17]

We have thus found support within halakhic interpretation for the two directions we proposed in Maimonides and Nachmanides. According to Maimonides, midrashic analogy does not rest upon induction, similar to what we saw in Rabbi Shimshon's view, whereas according to Nachmanides, the fundamental inference that underlies analogy is induction, similar to the Ra'avad's position.

It should be noted that I do not intend to base Maimonides' position on Rabbi Shimshon's view, but only to illustrate it by means of that view. According to Rabbi Shimshon, 'binyan av from two verses' is induction, and therefore, according to our line of thought, a conclusion derived from such a 'binyan av' is explicitly present in the text and ought, even according to Maimonides, to have the status of a Torah-level law. Yet Maimonides writes explicitly regarding all the rules by which the Torah is expounded that the laws derived from them have rabbinic force.

If so, there are three approaches: 1. The Ra'avad holds that even analogy is founded on induction, that is, induction ('binyan av') is the fundamental inferential method. 2. Rabbi Shimshon understands that both methods of inference exist in halakhic derash ('binyan av' and 'mah matzinu'), and neither is based on the other. 3. Maimonides maintains that analogy is the fundamental inferential method, and that it also underlies induction. 'Binyan av from two verses' is indeed induction, but induction, according to Maimonides' conception, is nothing more than a composition of analogies. Therefore, according to Maimonides, the status of all the methods of derash is like the principled status of analogy itself—namely, an extension of the text beyond what is contained in it. Consequently, according to Maimonides, the laws that arise from all the methods of derash have rabbinic force.

E. The Halakhic Status of Midrashic Laws According to Maimonides

Up to this point we have discussed the logical status of the methods of halakhic derash. I now wish to discuss briefly a more popular topic, namely the halakhic status of midrashic laws in Maimonides' system. As noted, many have dealt with this topic: some have held that their status is that of Torah-level laws, others that their status is that of rabbinic laws. Some have proposed an intermediate status called 'Divrei Sofrim,' and have suggested different characteristics for it.

As noted, textual considerations indicate that Maimonides' intention is to say that these are literally rabbinic laws, and this is not the place to elaborate. Yet even so, it is difficult to understand the status of these laws as literally rabbinic laws.

Nachmanides, in his glosses to the second root, asks Maimonides what the difference is between genuine derivations and those classified as mere asmakhta; apparently both are scriptural supports for rabbinic laws. Beyond that, it is ultimately clear that God's will is indeed expressed in these commandments. These laws are learned by means of methods that themselves are a law given to Moses at Sinai (the hermeneutic rules), from the text that was given at Sinai. If so, one cannot treat these derivations as ordinary rabbinic laws pure and simple.

Ordinary rabbinic laws are created on the basis of motives that exist among the Sages who enact them (a fence, a decree, an enactment, and so forth). Midrashic laws are not created from motives arising in the Sages' own reasoning, even according to Maimonides, but as interpretations of Scripture. The reason for the commandment to revere the sages is not the reasoning of the Sages as legislators, for otherwise this would simply be a rabbinic enactment and the derash would be an asmakhta. The reason for the obligation is an extension of the meaning of the rule written in the Torah, not a law with an independent rationale. Were it not for the derash on the direct-object marker, which serves to include Torah scholars, this law would not have been established in this way, but at most as an independent rabbinic enactment. Briefly put, a rabbinic law is a law created on the initiative of the Sages, whereas the midrashic law is imposed upon them by the text—that is, it is the 'initiative' of the Holy One, blessed be He.[18]

One may connect this intermediate status of midrashic laws with the status of laws given to Moses at Sinai. As noted, both of these types of law are manifestly God's will, and one may also say that they were transmitted de facto at Sinai. Yet these laws are not found in the text itself. That is precisely the root of the intermediate status given to them in Maimonides' doctrine.

For Nachmanides, as for most halakhic decisors, there are two levels of legal force in commandments: 1. Torah-level: laws transmitted at Sinai, or derived from what was transmitted there. 2. Rabbinic: laws that were innovated over the generations.

By contrast, in Maimonides' view there seem to be three levels: 1. Torah-level: laws that appear explicitly in the text. 2. Divrei Sofrim: laws that are God's will but are not explicit in the text. 3. Rabbinic: laws that are commands of the Sages on their own initiative (enactments, decrees, and so forth).

According to this, it is clear that derivations defined as asmakhtot are in fact ordinary rabbinic laws.[19] They belong to the third halakhic category in Maimonides' system. By contrast, ordinary midrashic laws, which are also rabbinic, belong to the second category (which we called above 'Divrei Sofrim'). It follows that according to Maimonides there are also three kinds of 'midrashic' laws in the Talmud: derivations supported by tradition—laws of the first type (Torah-level); ordinary derivations—laws of the second type ('Divrei Sofrim'); asmakhtot—laws of the third type (rabbinic).[20]

Much could be discussed regarding the halakhic implications of this division, but this is not the place for it.[21]

In conclusion, let us examine the expression Maimonides uses to describe his approach to halakhic midrashim: like branches issuing from the roots. This expression recalls, not accidentally, the aggadic teaching in the Talmud on the verse in Ecclesiastes 12, 'The words of the wise are like goads, and like well-planted nails…' In Babylonian Talmud, Hagigah 3b, Rabbi Yehoshua expounds this verse as follows:

If you would say: just as a nail diminishes and does not increase, so too the words of Torah diminish and do not increase—Scripture therefore says 'planted.' Just as a planting grows and multiplies, so too the words of Torah grow and multiply.

The words of the Sages can be described as causing the Torah to grow and multiply only according to Maimonides' conception as presented here. According to the accepted approach, the words of the Sages either expose what is already present in the Torah (through derivations), or add to it laws that are not derived from it (enactments and rabbinic laws). Neither of these actions constitutes the growth and multiplication of the words of Torah themselves. Only according to Maimonides does the midrash cause the Torah to grow and multiply—that is, it adds to it laws that were not in it: 'like branches'; and at the same time it does so in the manner of growth and multiplication, that is, it brings those branches forth from the Torah itself: 'from the roots.' The results of this process are laws that are in the category of 'the words of the Sages' (= rabbinic laws), but which nevertheless count as the growth and multiplication of the Torah itself, like nails that are planted within it.

[1] We saw there that this relation can be expressed as follows: analogy = induction + deduction.

[2] There is an extensive and detailed discussion of Maimonides' intention in these remarks. Does he mean that these laws are rabbinic, or Torah-level, or some intermediate state? See at length Neubauer, Maimonides on Divrei Sofrim (hereafter: Neubauer), and Rabbi Nahum Eliezer Rabinovitch's article, 'On Divrei Sofrim Whose Force Is Torah-Level in Maimonides' Doctrine,' Sefer Higgayon, Zomet Institute, Gush Etzion, 1995, p. 87. See also Rabbi Yeruham Fishel Perla's introduction to his edition of Saadia Gaon's Sefer HaMitzvot, in the section dealing with Maimonides' second root, and more.

Below I will assume that Maimonides' intention is to say that these are literally rabbinic laws, or at the very least not ordinary Torah-level laws. This is well anchored in his wording in the Arabic original of the roots. On this point see Moshe Halbertal, 'Maimonides' Sefer HaMitzvot,' Tarbiz 59, 1990, note 12 (hereafter: Halbertal). See also a brief discussion below at the end of our remarks.

[3] Admittedly, in the introduction to the Commentary on the Mishnah and also in the second root, the determination that a law given to Moses at Sinai has rabbinic force is not mentioned explicitly. However, elsewhere in his writings Maimonides indeed explicitly adopts such a position. See, for example, in the Freimann edition, responsum 161 (Blau, responsum 355). This responsum is cited by Nachmanides in his glosses to this root (in the Frankel edition, p. 54). This responsum is central for understanding Maimonides' method, but this is not the place to elaborate.

The determination that a law given to Moses at Sinai is a law with rabbinic force is also found in Maimonides' commentary to Mishnah Kelim 17:12.

[4] It is possible that these very problems led Maimonides to his revolutionary innovations in this topic.

[5] What causes confusion in understanding Maimonides' conception is his statement in the introduction to the Mishnah that laws transmitted by tradition from Moses at Sinai were never subject to dispute (see Havvot Yair, responsum 192; the point is well known). According to our discussion here, this does not mean that these are Torah-level laws. Laws given to Moses at Sinai have rabbinic status, but no dispute arose concerning them.

[6] A detailed discussion of this topic requires us to address the question of the relation between scriptural midrash and the plain meaning of the text. On this issue see David Henshke's two articles, 'A Verse Does Not Depart from Its Plain Meaning,' HaMa'ayan 17, Jerusalem, 1977, and Rabbi Witman's response in HaMa'ayan 18 (1978), together with Henshke's supplement there. In our discussion below, the positions of Maimonides and Nachmanides on this topic will become clear, and it will emerge that neither of them fits Henshke's interesting proposal there. Problems raised in Rabbi Witman's article are also resolved in light of the explanation I will offer below of Maimonides' view (see, for example, his note 1 on p. 49 regarding the term 'plain meaning by way of derash,' which he found difficult to understand).

[7] See, for example, Menachem Elon, HaMishpat HaIvri, and elsewhere.

[8] At the end of the day, one may understand this proposal as an explanation of Halbertal's distinction. Yet below I will try to show that this form of presentation has far-reaching implications for understanding the nature of the methods of derash in general, and that at its foundation lies a position different from Halbertal's regarding the understanding of the notion 'Torah-level' in Maimonides' doctrine.

[9] For a generalization of this principle, see the beginning of chapter 2 of the first gate in my aforementioned book.

[10] See, for example, Introduction to Logic, Hugo Bergman, Bialik Institute, Jerusalem, 1975, pp. 334, 349. On this subject in greater detail: Edmund Husserl, Cartesian Meditations, Magnes, Jerusalem, 1991, and the translator A. Z. Brown's introduction. Husserl is usually understood as proposing generalizations that do not rest on induction, and not as he is presented here, namely as providing a mechanism that grounds induction, but this is not the place for that.

[11] It is important to note that the commandment of awe toward God and toward the sages is brought here only by way of illustration. Nachmanides' position concerning the relation between these commandments is more complex, as can be seen in his glosses to the second root.

[12] See, for example, Kol HaNevuah in the aforementioned article, sections 16 and 17, especially the notes there, where he brings this from commentators and scholars.

[13] See Encyclopaedia Talmudit, vol. 4, entry 'Binyan Av.'

[14] In the Sifriyat Bnei Torah edition. The original edition does not appear in the copy in my possession. It is noted there that this printing is identical to the Amsterdam 1709 edition (with correction of many printing errors).

[15] See Encyclopaedia Talmudit, ibid. These sources are cited in Kol HaNevuah, p. 87, note 44; see there.

[16] A similar procedure is accepted in the methodology of modern science. A scientific theory (a general law of nature) cannot be proved, and therefore one arrives at it by induction from particular cases. The more numerous the particular cases that accumulate, the greater the confirmation of the general theory. As argued here, in halakhic derash as well we need at least two different cases that share a common feature in order to infer from them a general law.

[17] What is said here involves a certain novelty. The prevalent conception is that the mutual necessity between two source cases in a 'binyan av' ('what is common to them') appears in order to avoid an obstacle to the derivation because of the rule 'two verses that come as one do not teach.' Here I suggest that the mutual necessity is required for the derivation itself, so that the expounder can generalize inductively from those two cases to the relevant and correct general feature.

[18] In the laws of the Sabbath there is a well-known distinction between rabbinically prohibited labors and enactments, fences, and the like. The distinction here is similar.

[19] It would seem that asmakhtot must be defective derivations, for otherwise such derivations would create either Torah-level laws or ordinary 'Divrei Sofrim.' For comparison: according to Ralbag, asmakhtot are derivations that do not support known laws.

[20] I note briefly here that from the Ritva's interpretation in his novellae to tractate Rosh Hashanah of the notion 'asmakhta,' it emerges that even derivations that are merely asmakhta are not literally rabbinic laws created on the initiative of the Sages. According to the Ritva there, such laws have some root in the text, and in that sense they resemble midrashic laws as understood by Maimonides. See Encyclopaedia Talmudit, entry 'Asmakhta.'

[21] For example, one might think that doubt regarding midrashic laws should be ruled stringently, even though they are rabbinic laws, since after all there is here a concern about violating God's will, or about the spiritual damage that the commandment is intended to prevent.

This discussion is especially complicated in Maimonides' own view, since he holds that a Torah-level doubt is treated leniently by Torah law. Regarding doubt in a law given to Moses at Sinai, see, for example, Mishneh Torah, Hilkhot Shehitah 5:3, and the notes of the commentaries there. Some maintain that its doubt is treated leniently even according to those who hold that it is Torah-level (see, for example, Encyclopaedia Talmudit, entry 'Halakhah le-Moshe mi-Sinai').