Lighting Shabbat Candles in Jerusalem 40 Minutes Before Sunset
By William Gewirtz
There is a story, perhaps apocryphal, of a visit to Jerusalem by R. Yoel Teitelbaum in which he is driven to the Kotel on Friday afternoon well after the customary time to light Shabbat candles in Jerusalem, 40 minutes before sunset. As his car was being stoned, he suggested that instead of adding 40 minutes to the Friday night pre-Shabbat period, it would be more appropriate that 40 minutes be added to the time at which the calendar of Jerusalem announces that Shabbat ends. That he had little regard for the ancient customs of Jerusalem is probably not surprising; finding a compelling rationale for the zemanim practiced in Jerusalem is a wholly other matter. In terms of Saturday evening, Jerusalem has always followed the opinion of the geonim, which is now most often attributed to the Gaon of Vilna. For the entire period of recorded history, even prior to the era of the Gaon, with isolated exceptions, Shabbat ended in Jerusalem at most 36 to 42 minutes after sunset, depending on the season. However, some returning from Europe brought back with them to Israel the European practice that extended Shabbat to 72 minutes after sunset or even further in accordance with the opinion of Rabbeinu Tam.
However, lighting candles 40 minutes before sunset Friday night remains baffling despite several theories that have attempted to provide a rationale, all of whom I find questionable. Why 40 minutes instead of 18, 20, 30, 36 or 45 minutes? What follows are halakhic positions from authorities going back over 800 years, and perhaps even supported by a source in the yerushalmi, which provides a theory that is consistent with practices rarely encountered in recent times. As we will see, many of these practices must contend with issues that cannot be defended in their entirety without some minor modification / correction. Ironically, the standard alternative often observed, based on Ramban and many subsequent ḥakhmai sforad, also faces a major issue that I cannot effectively address.
What follows are an organized sequence of ten propositions that provides clear support for the practice of lighting candles 40 minutes before sunset; I succinctly demonstrate clear support for each proposition from major sources and / or figures in halakhic history. Despite its formal organization, this essay presents an educated guess as opposed to a definitive conclusion. In other contexts, I have warned against being overwhelmed by numerical coincidences; though I strongly doubt it, one cannot rule out that this is just another example of one as well.
Deriving 40 minutes before sunset
Proposition 1. The hours of the day were separately estimated from a morning start point to midday and from midday to an evening endpoint. It is highly unlikely that calculating the length of time between a morning start point until an evening endpoint and dividing by 12 was used in that manner to determine the length of a halakhic hour prior to the existence of clocks.
Support: While calculating from a point in the morning to a point in the evening and dividing by twelve is the theoretical method implied in the Talmud, it seems rather unlikely to have been used in practice prior to the benefit of a clock. In fact, in describing his method of estimation of the time by which to finish the consumption of ḥametz on erev pesaḥ, Ravyah explicitly describes his method for estimating the morning hours between a morning start point and ḥatzot. This assumption about separately calculating from ḥatzot to both a morning and evening endpoint is critical to what is proposed in this essay.
Proposition 2. The morning start point used in the Middle East was alot ha’shaḥar, not sunrise, despite the influence of the talmidei ha’gra. In addition, in Jerusalem, 90 versus 72 minutes before sunrise was often, but not always, the time used for alot ha’shaḥar around the fall and spring equinox.
Support: Using alot ha’shaḥar as the morning start point is rooted in the opinions of Ramban, R. Israel Isserlein and many other rishonim. Clearly, the Ben Ish Ḥai and the calendar of Jerusalem, among many others, calculated using alot ha’shaḥar versus sunrise. The use of 90 versus 72 minutes before sunrise as the time of alot ha’shaḥar occured at various times in history in Eretz Yisroel and other parts of the Middle East as well, particularly in Jerusalem. Whether the Gaon supported 90 or 72 minutes is strongly disputed.
Proposition 3. The evening endpoint is either the symmetric counterpoint to alot ha’shaḥar, as is clearly derivable from Ramban and his school, or an asymmetric point in the evening occurring significantly earlier at the point of transition between days of the week according to the geonim. Finding support from R. Israel Isserlein for such asymmetric endpoints is a complicated and debatable task that is, in any case, arduous to demonstrate. Instead, we reference explicit support from multiple significant aḥaronim.
Support: Clearly Ramban and his school who assert that plag ha’minḥa occurs only 3.75 minutes before sunset were calculating from a point as far after sunset as alot ha’shaḥar is before sunrise. Astounding as it might seem, numerous important aḥaronim calculated to an asymmetric earlier endpoint, approximately 20-40 minutes after sunset. Among aḥaronim who maintain such a viewpoint are R. Nosson Adler, R. Yaacov Lorberbaum, the Ben Ish Chai, R. Yosef Chaim Sonnenfeld, among many others. While the endpoint of Ramban is the point at which Shabbat ends according to Rabbeinu Tam, the earlier point is the end of Shabbat according to the geonim.
Support for such asymmetry can also be derived from a lengthy (and disputed) discussion beginning in yerushalmi Berakhot 2b, that considers the verse in Nehemiah 4:15,
- Ve’anaḥnu osim be’melaḥah…. Mei’alot ha’shaḥar ad tzait ha’kokhavim, as defining an asymmetric daytime period from alot ha-shaḥar until the appearance of three stars.
Proposition 4. With clocks in common use, each of the aḥaronim mentioned counted the length of time from alot ha’shaḥar to an earlier evening endpoint and divided by 12 to derive the length of a halakhic hour. This method of calculation resulted in the miscalculation of ḥatzot.
Support: Their method is an unarguable fact that appears in their writings and / or calendars. One can also easily verify the miscalculation of ḥatzot by calculating a halakhic hour using alot ha’shaḥar and end of Shabbat according to the geonim as endpoints. That calculated point of ḥatzot is typically 20-30 minutes earlier than the indisputable point of ḥatzot that can be observed directly.
Proposition 5. Because of a miscalculated ḥatzot, some wanted to throw out the baby with the bathwater and claimed the absolute necessity of using Ramban’s later endpoint that is symmetric with alot ha’shaḥar. Use of any symmetric endpoints around sunrise and sunset calculates the point of ḥatzot correctly.
Support: The most complete account of this issue and its ramifications come from various documents recording the debate that took place over several years in Jerusalem more than 110 years ago between R. Yosef Chaim Sonnenfeld and R. Yeḥiel Miḥel Tukatzinsky. The calendar originally in use, strongly supported by R. Sonnenfeld, miscalculated ḥatzot. Multiple insignificantly different accounts of the debate all agree that in the end a changed calendar that calculated ḥatzot accurately resulted. By moving ḥatzot forward by about 20 minutes, the new calendar also set sof zeman keriat shema about 10 minutes later, which was the primary motivation for R. Sonnenfeld’s objection. The changed calendar, like the current calendar (still) in use today, calculates using a depression angle of approximately 20 degrees, equivalent to 90 minutes around both the spring and fall equinox, identical to what Ramban proposed.
Proposition 6. Unfortunately, Ramban’s endpoints, 90 minutes from sunrise and sunset when used around the winter solstice, results in plag ha’minḥa occurring around 10 minutes after sunset, an inelegant and disqualifying occurrence. The fact that this has not been recognized would imply that this version of the opinion of Magen Avraham using 90 minutes, with either a fixed or a depression angle implied number of minutes, was not in widespread use.
Support: This is indisputable if we examine dates near the winter solstice. On December 21 in Jerusalem, sunset is at 6:39 PM and plag ha’minḥa occurs between 7 and 13 minutes after sunset depending if you calculate with a fixed 90 minutes (strongly opposed but resulting in 8 minutes) or depression angles (strongly supported and resulting in 13 minutes.)Note that 72 minutes does not have this problem; plag ha’minḥa occurs very slightly before sunset on December 21 when 72 minutes is used.
Proposition 7. Fixing the alternative that miscalculates ḥatzot is straightforward; just calculate like we assume occurred before the use of clocks - from a known point of ḥatzot to alot ha’shaḥar and from ḥatzot to an earlier evening endpoint. Note that ḥatzot is not calculated but observed and occurs at midday.
Support: The morning hours present no issues; find the length of time between alot ha’shaḥar and ḥatzot and divide by six. Afternoon hours are a bit stickier. There are multiple options for the precise time to use for the evening endpoint, depending on one’s best estimate of the point of transition between days of the week on a biblical level. One could advance arguments for any depression angle that associates with a time between 20 - 28 minutes after sunset around the spring and fall equinox. Given the preference for 90 minutes over 72 in Jerusalem, use of such an earlier endpoint, which avoids the (unreported and) anomalous occurrence of plag ha’minḥa after sunset, appears to be reasonable.
Proposition 8. Those who note that morning hours are longer than afternoon hours need not be concerned; in an unexplained position, one of last century’s greatest poskim claimed that unequal morning and afternoon hours is not an anomaly but what should be expected.
Support: In a position that neither I nor the many who I have asked can fully explain, R. Moshe Feinstein insisted that halakhic hours differ between the afternoon and the morning. Unfortunately, R. Feinstein states that either the morning or afternoon hours can be longer; this approach can only explain the morning hours being longer. While I cannot claim that this approach provides the definitive explanation, I have never found another approach that provides any more cogent (albeit partial) rationale.
Proposition 9. Using this approach or even the errored one that miscalculates ḥatzot, find the time of the year when plag ha’minḥa comes closest to sunset.
Support: The time for plag ha’minḥa comes closest to sunset around December 21st when the daytime period and hence halakhic hours are shortest. There are multiple opinions that differ slightly with respect to the biblical point after sunset that marks the transition between days of the week. Using a depression angle of 6 degrees, a reasonable choice for that point of transition, on December 21st plag ha’minḥa occurs 42 minutes before sunset. Throughout the rest of the year plag ha’minḥa occurs more than 42 minutes before sunset. Examining the issue in detail and using December 21st:
- sunset at 4:39 PM, and
- a depression angle of 6 degrees as the day’s approximate end, 27 minutes after sunset at 5:06 PM,
- a halakhic hour of ((ḥatzot to sunset) + 27 minutes) / 6 = (302 minutes + 27 minutes) / 6 = 54.833 minutes, resulting in
It is unimaginable that such a precise calculation that results in plag ha’minḥa 42 minutes before sunset was used to initially establish the custom of lighting 40 minutes before sunset. Additionally, many potential changes including:
- calculating (incorrectly) from alot ha’shaḥar,
- choosing a slightly earlier (or even (incorrectly) a later) evening endpoint,
- not using depression angles (an absolute certainty), and
- disagreements about how shekiah is to be calculated given Jerusalem’s altitude
will move the time of plag ha’minḥa, most often several minutes earlier.
However, it is critical to appreciate that we are attempting based on (halakhically inspired) religious instincts to light candles as early as is possible without violating an explicit halakhic boundary that demands that we light candles after plag ha’minḥa. Any attempt to light earlier than 40 minutes before sunset would likely face halakhic resistance, particularly at a time when estimation and approximation were still in common use.
Proposition 10. Lighting candles 40 minutes before sunset guarantees we are lighting at:
- a uniform time all year,
- as early as possible, but
- always at a time that is after plag ha’minḥa.
Support: 40 minutes is the largest round number that simultaneously meets all three proposed objectives. Q.E.D.
To again be clear, I do not claim that the original basis was derived as I have outlined. Undoubtedly, the original custom resulted from accurate approximation as opposed to precise calculation. Nonetheless, proposition 10 likely captures the original intent of those who started this unique practice. Knowing more of the early history surrounding this well establish custom would add significantly to our understanding. For now, it remains a conjecture on which comments would be appreciated.
 Even Hazon Ish waited only 45 minutes before ending Shabbat.
 See Minhagei Yisrael (page 102, footnote 18) by R. Yaacov Gliss and Ha’zemanim Ka’halakha (chapter 60, footnote 18) by R. Chaim Benish for proposed theories.
 While most currently follow the method of the Gaon of Vilna and calculate from sunrise to sunset, surprisingly, this method has no uncontested support prior to the 16th century when it was suggested by R. Mordechai Yaffe. Both R. Yaffe and the Gaon cited no prior halakhic support; instead they claimed that the hours of the day are naturally defined by the period between sunrise and sunset. This contentious topic is not pursued further.
 Multiple comments on different sections of the Shulḥan Arukh strongly imply support for 90 minutes; some comments in midrashic settings explicitly support 72 minutes.
 A student of R. Yisroel Isserlein, R. Yaacov ben Moshe in his sefer Leket Yosher sheds light on this issue, (assuming knowledge of the operation of the diverse clocks in use during the 15th century.) In the first mention of clocks in halakhic literature around the turn of the 16th century, R. Yaacov ben Moshe specifies that the time that R. Isserlein permitted a person having difficulty fasting on Taanit Esther to read the Megillah as slightly before 5 PM. What R. Isserlein described halakhically as plag ha’minḥa was quantified by R. Yaacov ben Moshe as occurring a few minutes before 5 PM.
 The perhaps unfamiliar 3.75 minutes is 1/6th of the time to walk a mil of 22.5 minutes.
 Ramban in Torat ha’Adam states that plag ha’minḥa occurs at the time it takes to walk 1/6th of a mil before sunset. From that statement three conclusions can be drawn:
- The time to walk a mil is 22.5 minutes, not the normally assumed 18 minutes.
- The hours of the day are calculated between alot ha’shaḥar and an evening equivalent, following what is referred to currently as the position of Magen Avraham.
- Alot ha’shaḥar and its evening equivalent are separated from sunrise and sunset respectively by 90 (not 72) minutes around the spring and fall equinox.
 R. Adler’s practice is still followed in Zurich.
 An abbreviated discussion also occurs in multiple places in the bavli.
 Three stars appear after sunset in the Middle East before 30 minutes after sunset. The Gaon of Vilna succinctly and accurately describes his view of the point of transition between days of the week as the appearance of 3 stars versus Rabbeinu Tam’s view that he equates to the appearance of “all the (millions of) stars.
 As traditional a posek as R. Yitzchok Weiss, the author of Minhat Yitzhak (vol 4:53), invalidates any approach that results in a miscalculation of hatzot.
 A young man at the time, R. Tukatzinsky was married to the granddaughter of the venerable R. Shmuel Salant, the last undisputed chief rabbi of Jerusalem in whose court the dispute was adjudicated.
 The length of the day on December 21st is 10 hours and 4 minutes. Using fixed minutes thus adding 180 minutes, dividing by 12 and multiplying by 1.25, (604 + 180)/12*1.25 = ~ 82 minutes, which puts plag ha’minḥa 8 minutes after sunset. Adding 192 versus 180 minutes results in plag ha’minḥa occurring another 5 minutes later, 13 minutes after sunset.
 All the morning hours, including sof zeman krait shema are identical to the hours calculated by any symmetric calculation based of the Magen Avraham’s opinion, as should be obvious and, in any case, easily verified.
 Related perhaps, but in ways that are unclear, both of last century’s most noted poskim, R. Shlomo Zalman Auerbach as well as R. Feinstein issued rulings about ḥatzot ha’lailah and ḥatzot ha’yom, respectively, that are incredulous. R. Feinstein writes based on tradition, but with no additional justification, that ḥatzot is not calculated and at the same time all years long. That ḥatzot is not calculated comports with the ancient practice illustrated by Ravyah that the determination of ḥatzot does not involve calculation but only observation; the latter, that ḥatzot occurs at same time all year long, remains unexplained. R. Auerbach’s ruling, which calculates ḥatzot ha’lailah for purposes of the pesaḥ seder, is yet more perplexing. That both poskim have baffling positions in approximately the same area, both of which have not been definitively explained, is intriguing.
 I am distinguishing religious from halakhic similarly to their different meanings as occur in the writings of both R. Joseph Soloveitchik and Prof. Jacob Katz.