ELONIOELONIO
ELONIO
Mission—Team—FAQ—Privacy—Terms—Sponsors—Grammar—Typing Course—Donate
© 2026 Elon Automation B.V.
  1. Grammar
  2. /Hebrew Grammar
  3. /Proverbs & Sayings
  4. /Mishnah: אל תסתכל בקנקן אלא במה שיש בו

Mishnah: אל תסתכל בקנקן אלא במה שיש בו

אל תסתכל בקנקן אלא במה שיש בו — "don't look at the jar, but at what is in it" — is quoted in Israel the way "don't judge a book by its cover" is quoted in English, and almost always on its own. It is in fact the last clause of a composite mishnah, and it is a rebuttal: two earlier speakers have just argued about whether you should learn from the young or from the old, and this line closes the argument by refusing the terms of the question. Read without the setup, the jars come out of nowhere.

The text is Mishnah, Avot 4:20 — around 200 CE, public domain, quoted here entire from the Torat Emet text on Sefaria. Some editions divide chapter 4 differently and number this mishnah 4:27. The question marks below are this page's; the printed mishnah has commas. The bare form of the quotations is ktiv haser, so ניר below is modern נייר and שאפלו is modern שאפילו; everything outside the quotations is ordinary ktiv male.

The whole mishnah

אלישע בן אבויה אומר, הלומד ילד למה הוא דומה? לדיו כתובה על ניר חדש.

elishá ben avuyá omér, haloméd yéled lemá hu domé? lidyó khetuvá al neyár khadásh.

Elisha ben Avuya says: One who learns as a child — what is he like? Like ink written on fresh paper. (Mishnah, Avot 4:20; the edition leaves the כ of כתובה spirantized after the vowel of לדיו, so it is read khetuvá — an Israeli quoting the line casually says ktuvá)

והלומד זקן למה הוא דומה? לדיו כתובה על ניר מחוק.

vehaloméd zakén lemá hu domé? lidyó khetuvá al neyár makhúk.

And one who learns when old — what is he like? Like ink written on erased paper.

רבי יוסי בר יהודה איש כפר הבבלי אומר, הלומד מן הקטנים למה הוא דומה? לאכל ענבים קהות ושותה יין מגתו.

rabí yoséi bar yehudá ish kfar habavlí omér, haloméd min haktaním lemá hu domé? le'okhél anavím kehót veshoté yáyin migitó.

Rabbi Yosei bar Yehuda of Kfar HaBavli says: One who learns from the young — what is he like? Like someone eating unripe grapes and drinking wine straight from the press.

והלומד מן הזקנים למה הוא דומה? לאכל ענבים בשלות ושותה יין ישן.

vehaloméd min hazkením lemá hu domé? le'okhél anavím beshelót veshoté yáyin yashán.

And one who learns from the old — what is he like? Like someone eating ripe grapes and drinking aged wine.

רבי אומר, אל תסתכל בקנקן, אלא במה שיש בו.

rabí omér, al tistakél bakankán, éla bemá sheyésh bo.

Rabbi says: Do not look at the jar, but at what is in it.

יש קנקן חדש מלא ישן, וישן שאפלו חדש אין בו.

yesh kankán khadásh malé yashán, veyashán she'afílu khadásh ein bo.

There is a new jar full of old wine, and an old one with not even new wine in it.

Who says it — a genuine textual split

Printed editions of the Mishnah introduce the last speaker as רַבִּי with no further name, and in the Mishnah רבי unqualified means Rabbi Yehuda HaNasi, the tractate's own redactor. But that is not the only reading.

WitnessReadsSpeaker
Vilna printed Mishnah (Romm, 1913); Torat Emetרַבִּי אוֹמֵר rabí omérRabbi Yehuda HaNasi
Kaufmann manuscript A50, ed. Dan Be'eriרְבִּי מֵאִיר אוֹמֵר rebí me'ír omérRabbi Meir

The Kaufmann manuscript is generally taken as the best single witness to the text of the Mishnah, and it is not alone here: Avigdor Shinan's commentary on the mishnah — Pirkei Avot: A New Israeli Commentary, Jerusalem, 2009, where the mishnah is numbered 4:27 — notes flatly that בנוסחים רבים מיוחס המאמר לר׳ מאיר ולא לרבי יהודה הנשיא — "in many textual witnesses the saying is attributed to R. Meir and not to Rabbi Yehuda HaNasi." So the very common English attribution to Rabbi Meir is not a popular error to be corrected; it reflects a real manuscript tradition. The honest statement is that the printed text says רבי and the manuscript tradition is divided.

💡
Notice what the same manuscript does to the noun: Kaufmann vocalizes בַּקִּנְקֵן bakinkén where the printed text has בַּקַּנְקַן bakankán. A vocalization is an editor's reading of an unpointed consonantal text, so wherever two editions point differently they are telling you the consonants underdetermine the word — which is the ordinary condition of written Hebrew, not an exotic one.

אל + future: the negative command

Hebrew's prohibitive is אַל plus the future form. It is not the infinitive, not a special mood, and not the imperative negated — the imperative simply has no negative in Hebrew.

אל תשכח לכבות את האור לפני שאתה יוצא

al tishkákh lekhabót et ha'ór lifnéi she'atá yotsé

Don't forget to turn the light off before you leave. (to a man)

אל תדאגי, כבר שלחתי להם את המסמכים

al tid'agí, kvar shalákhti lahém et hamismakhím

Don't worry, I've already sent them the documents. (to a woman)

Now the part that matters. לא + future is not simply the wrong way to do this — it is a different sentence, and which sentence depends on which layer of Hebrew you are in:

ShapeLayerReads asExample
אל + futureclassical and modern alikeprohibitionאל תסתכל — don't look
לא + futuremodern Hebrewstatement or predictionלא תסתכל — you won't look
לא + imperfectBiblical Hebrewapodictic law — thou shalt notלא תגנב — you shall not steal

לא תגנב

lo tignóv

'You shall not steal.' (Exodus 20:13 in the Hebrew verse division, 20:15 in most English Bibles — the Biblical layer, where לא + imperfect IS the prohibition)

That third row is why the second one feels loaded. A modern Israeli who hears לא תסתכל in a commanding tone hears the Decalogue register, not an ordinary instruction — the form is available, it just carries the wrong weight. The everyday prohibitive is אל, and it has its own page at אל + future.

תסתכל, and the two things that happen to the ת

Hitpa'el is built by prefixing הִתְ־ to the root. Do that to ס־כ־ל and you would expect התסכל. Nobody says that, and it is not a word. What the language actually produces is *הִסְתַּכֵּל — the ת has moved behind the sibilant.

This is where almost every learner loses a hitpa'el verb in the dictionary, because the root's first letter is no longer where the prefix left it. And it is worth being precise, because two different things are going on and most descriptions merge them.

Process one — metathesis. The ת and the sibilant swap places. Nothing else changes. This happens after ס, שׂ and שׁ.

Process two — metathesis and assimilation. After צ and ז the ת also changes quality, because it now sits next to a consonant it cannot match: emphatic צ pulls the ת to emphatic ט, and voiced ז pulls it to voiced ד.

RootHitpa'elReadsWhat became of the ת
ס־כ־להִסְתַּכֵּלhistakélmoved behind the ס
ס־ד־רהִסְתַּדֵּרhistadérmoved behind the ס
שׁ־מ־שׁהִשְׁתַּמֵּשׁhishtaméshmoved behind the שׁ
שׂ־כ־רהִשְׂתַּכֵּרhistakérmoved behind the שׂ
צ־ר־ףהִצְטָרֵףhitstaréfmoved and became ט
צ־ע־רהִצְטַעֵרhitsta'érmoved and became ט
ז־ק־ןהִזְדַּקֵּןhizdakénmoved and became ד
ז־מ־ןהִזְדַּמֵּןhizdaménmoved and became ד

Neither process is optional and neither is historical residue: form a hitpa'el on any sibilant-initial root and it comes out this way automatically. That is also the practical rule for the dictionary — delete the ה, then delete the ת wherever it has ended up, counting any ט or ד as that same ת. What is left is the root, already in its own order: הסתכל files under ס־כ־ל, הצטרף under צ־ר־ף, הזדמן under ז־מ־ן. There is no entry under ה־ס־ת and never will be. The full paradigm is on hitpa'el metathesis.

השתמשתי בזה פעם אחת ומאז זה עומד בארון

hishtamáshti bazé pá'am akhát ume'áz ze oméd ba'arón

I used it once and it's been sitting in the cupboard ever since.

הצטערתי מיד אחרי ששלחתי את ההודעה

hitsta'árti miyád akharéi sheshalákhti et hahoda'á

I regretted it the moment I sent the message.

אלא, where אבל cannot go

אֶלָּא is a corrective conjunction: it requires a preceding negation and it replaces the negated element. אֲבָל is contrastive: it adds a clause that stands against the first one. English "but" does both jobs, which is why the pair is so hard to keep apart — and both of the sentences below are correct Hebrew saying different things.

לא ביקשתי הנחה אלא הסבר

lo bikáshti hanakhá éla hesbér

I didn't ask for a discount, I asked for an explanation. (אלא swaps in the right word for the negated one)

לא ביקשתי הנחה, אבל הם נתנו לי אחת בכל זאת

lo bikáshti hanakhá, avál hem natnú li akhát bekhól zot

I didn't ask for a discount, but they gave me one anyway. (אבל adds a contrasting clause instead of correcting the first)

In the mishnah, אל תסתכל בקנקן is the negation and במה שיש בו is what replaces בקנקן, so אלא is the only possibility. The pair, including לא רק … אלא גם and אלא אם כן, is laid out on אבל and אלא.

במה ש־, and the pronoun English throws away

מַה שֶּׁ־ is Hebrew's free relative — "what, that which" — and it takes its preposition on the outside: בְּמַה שֶּׁ־ "in what", עַל מַה שֶּׁ־ "about what", מִמַּה שֶּׁ־ "than what".

Inside the clause, something else is required. שֶׁ־ is a pure subordinator with no case of its own — unlike English which or whom, it cannot itself be the object of a preposition. So whenever the relativised element is anything other than a plain subject or direct object, its role has to be spelled out inside the clause with a resumptive pronoun. In the mishnah, that pronoun is בּוֹ: literally at what that there is in it.

הדירה שגרנו בה הייתה קטנה בטירוף

hadirá shegárnu ba haytá ktaná betirúf

The flat we lived in was insanely small. (בה is obligatory — English deletes 'in it', Hebrew cannot)

זה בדיוק מה שדיברנו עליו אתמול

ze bediyúk ma shedibárnu aláv etmól

That's exactly what we talked about yesterday. (עליו — 'about it' — cannot be dropped)

הספר שקראתי בשבוע שעבר היה הרבה יותר טוב

haséfer shekaráti bashavúa she'avár hayá harbé yotér tov

The book I read last week was much better. (a direct object, so no resumptive pronoun is needed here)

Two jars, no nouns

The last line runs on יֵשׁ and אֵין, the existential pair, and on adjectives standing in for nouns that never appear. חָדָשׁ and יָשָׁן — "new" and "old" — are doing the work of "new wine" and "aged wine" with no noun in sight, and מָלֵא יָשָׁן takes its complement bare, with no equivalent of English full of. Modern Hebrew keeps all of it: הזקנים the old, הצעירים the young, החדש והישן the new and the old.

אין פה שום דבר חדש, זה בדיוק מה שאמרנו לפני שנה

ein po shum davár khadásh, ze bediyúk ma she'amárnu lifnéi shaná

There's nothing new here — it's exactly what we said a year ago.

Common Mistakes

1. Using לא to forbid something. English negates commands with don't, one word for all of it, so learners attach Hebrew's ordinary negator to a command. The result is grammatical and is not a command.

❌ לא תסתכל בטלפון בזמן שאני מדבר איתך

Incorrect as a prohibition — in modern Hebrew לא + future is a prediction, 'you won't look'. A command needs אל: אל תסתכל בטלפון.

✅ אל תסתכל בטלפון בזמן שאני מדבר איתך

al tistakél batélefon bizmán she'aní medabér itkhá

Don't look at your phone while I'm talking to you. (to a man)

2. Building hitpa'el straight off the template. You learn הִתְקַטֵּל, you know the root, and you assemble the two. On a sibilant root the language does not permit the result.

❌ התסכל, התצרף, התזמן

Incorrect — hitpa'el moves the ת behind a sibilant, and after צ and ז changes it as well: הסתכל, הצטרף, הזדמן.

✅ הצטרפתי לקבוצה החדשה, אבל אני כמעט לא מסתכל שם

hitstarápti lakvutsá hakhadashá, avál aní kim'át lo mistakél sham

I joined the new group, but I hardly ever look at it. (m. speaker — both processes in one sentence: הצטרפתי with ט, מסתכל with a plain ת)

3. Correcting a negation with אבל. English uses but for both jobs, so whichever Hebrew word you learned first gets used for both.

❌ אל תסתכל בקנקן אבל במה שיש בו

Incorrect — after a negation, 'but rather' is אלא. אבל joins two clauses in contrast; it cannot swap in a replacement for the negated word.

✅ זה לא באג אלא פיצ׳ר, כך לפחות אמרו לנו

ze lo bag éla fítsher, kakh lefakhót amrú lánu

It's not a bug, it's a feature — or so we were told.

4. Deleting the resumptive pronoun. English relative clauses drop the preposition's object routinely (what is in it becomes what is in, in Hebrew's eyes), and the Hebrew that results says something else.

❌ אל תסתכל בקנקן אלא במה שיש

Incorrect as 'at what is in it' — without בו the clause reads 'at what there is', and the jar has disappeared from the sentence. ש־ carries no case, so the role is spelled out inside: שיש בו.

✅ תשפוט אותו לפי מה שהוא עשה, לא לפי מה ששמעת עליו

tishpót otó lefí ma shehú asá, lo lefí ma sheshamáta aláv

Judge him by what he did, not by what you heard about him. (to a man — עליו is obligatory in the second clause and impossible in the first)

Key Takeaways

  • The line is the last clause of a composite mishnah and a rebuttal of the two speakers before it, not a free-standing aphorism.
  • The speaker is contested by the manuscripts: printed editions read רבי (Rabbi Yehuda HaNasi), the Kaufmann manuscript reads רבי מאיר, and Shinan reports many witnesses with Rabbi Meir. The common English attribution is a variant, not an error.
  • אל + future is the prohibitive. לא + future is a prediction in modern Hebrew and apodictic law in Biblical Hebrew — one shape, three readings by layer.
  • Hitpa'el does two separate things behind a sibilant: it reorders the ת after ס, שׂ, שׁ, and it reorders and changes it after צ (giving ט) and ז (giving ד). Undo both to find the root in a dictionary.
  • אלא replaces a negated element; אבל contrasts two clauses. After a negation only אלא can do the correcting job.
  • ש־ has no case, so anything beyond a subject or direct object needs a resumptive pronoun inside the clause: שיש בו, שגרנו בה, שדיברנו עליו.

Related Topics

  • אל: The Negative CommandA2 — To tell somebody not to do something, Hebrew uses אל plus a second-person future — אל תלך, never the imperative form itself. That is why לך and אל תלך look unrelated, and why the textbook line that לא plus the future is 'wrong' is itself wrong: it is a different register, not a different grammar.
  • When the ת Moves: הסתכל, השתמש, הצטלם, הזדקןB1 — If the first root letter is a sibilant, the ת of the hitpa'el prefix changes places with it — and next to צ it hardens to ט, next to ז it voices to ד. Three outcomes across five letters, one phonetic reason, and a separate mechanism that is routinely confused with it.
  • אבל and אלא: Two Kinds of 'But'A2 — English has one but and Hebrew has two, and the split is not a matter of style: אבל adds a contrasting clause, אלא swaps a correct item in for one you have just rejected. English speakers do not notice the second slot exists, which is why the calque with אבל is the most reliable tell there is.
  • מה ש… and מי ש…: Free RelativesB2 — Hebrew turns question words into noun-like clauses with ש־: מי שרוצה ‘whoever wants,’ מה שקרה ‘what happened,’ and the wider family איפה ש, מתי ש and איך ש.
  • Hillel: אם אין אני לי, מי לי?B2 — Three conditional clauses at Mishnah Avot 1:14 with no verb in any of them — a close reading of verbless predication, of אין negating a pronoun subject directly, and of the inflected preposition ל that carries a whole predicate on its own, with the three live modern ways of saying I am not.
← PreviousHillel: אם אין אני לי, מי לי?Next →זה, זאת, אלה: This and These