From the Rif on Shabbat,
daf 5b (corresponding to Shabbat 12b)
One brayta said: An attendant may examine glasses and plates by the light of a lamp. And another brayta said: He must not examine [them].
This is not a difficulty.
One refers to a permanent attendant, the other to one who is not permanent.
And if you want I will say: both refer to a non-permanent attendant {our gemara: permanant} yet there is no difficulty: one refers to [a lamp fed with] oil, the other to naphtha {which emits an unpleasant odor, and produces a lot of light, and so he won't come to tilt it}.
It was a question to them: What of a permanent attendant {our gemara: non-permanent} and a [lamp fed with] oil?
Rav Huna {our gemara: Rav} said: It is the halacha and we do not teach so {to others}.
Rabbi Yirmiya bar Abba said: It is the halacha and we do teach so.
{The Rif concludes:} We learn now that whether with a permanent or a non-permanent attendant, he may inspect dishes and cups by the light of the lamp, and specifically using naphta, which is disgusting, and produces a lot of light, and he won't need to tilt it. But with oil, a non-permanent attendant for certain is forbidden, and regarding a permanent attendant they argue, and the halacha is that he may inspect, and we do not teach so.
As noted in the curly brackets {}, there are several differences between our version of the Gemara and that of the Rif. One fairly important one is the permanent/temporary attendant switchoff. Whereas the gemara as we read it has it that the permanent attendant as the one who we regard more stringently, the Rif has it that we deal more strictly with the temporary attendant.
We could come up with reasons why we should be more strict with regard to the temporary or permanent attendant. For the permanent one, we have Rashi's explanation that he fears his master and will therefore look more closely to ensure that there is no dirt on the tableware. For the temporary one, perhaps one can say that a permanent attendant knows his craft of cleaning/inspecting the tableware better, and would not need as much light, and will know not to, or will not come to tilt the lamp.
Which is the better reading? Well, the Rif's version solves a difficulty in the continuation of the gemara:
Rabbi Yirmiya bar Abba visited Rav Assi. Now, his attendant arose and examined [the glasses] by candlelight. Thereupon his [Rav Assi's] wife said to him [Rav Assi], 'But you do not act thus!' 'Let him be,' he answered her, 'he holds with his master {Rabbi Yirmiya bar Abba}.'
Now it truly seems from the story that this is Rabbi Yirmiya bar Abba's usual, and thus
permanent, attendant. Yet according to our Gemara, Rabbi Yirmiya's statement was restricted to a
temporary attendant, {even} using an oil lamp. So if Rabbi Yirmiyah's statement will apply {and it does, as we know from Rav Assi's response to his wife}, it must deal with a
temporary attendant.
Rashi gives the somewhat forced answer that since this is in someone else's home, he has the status of a temporary attendant, even though he is Rabbi Yirmiyah's usual attendant.
However, if you switch permanent with temporary {"not-permanent"} throughout the gemara's give and take {as in the Rif's version}, then Rabbi Yirmiyah's statement is on a
permanent attendant, who is allowed to do this even with an oil lamp, and we teach so, and so we would expect his permanent attendant to conduct himself exactly as his attendant does in the story.A temporary attendant would not be allowed to do so.
On the other hand, perhaps one could argue that our version is better. Consider the two
braytas:
One brayta said: An attendant may examine glasses and plates by the light of a lamp.
And another brayta said: He must not examine [them].
We might perhaps expect that a resolution would list the lenient case first and the strict case second, paralleling the order of the two
braytas. We do not have this consistently.
Here is what we have, in both our gemara and the citation of the Rif:
There is no difficulty: one refers to a permanent attendant, the other to a temporary one.
According to our gemara, the strict is mentioned first, and the lenient mentioned second. According to Rif, the lenient is mentioned first, and the strict mentioned second. Consider the continuation, though:
And if you want I will say: both refer to an X {the stricter one} yet there is no difficulty: one refers to [a lamp fed with] oil, the other to naphtha.
Here, the first case is the strict, and the second case is the lenient. It makes more sense for the strict/lenient order of the two suggestions to be consistent with each other, which is true for our girsa and not that of the Rif. On the other hand, I should look at Dikdukei Sofrim to see what other girsaot there are, and further, I should really compare other sugyot and see if the suggestions are listed there in consistent order, such that this is a legitimate expectation.
If the order of strict/lenient is expected to be consistent across suggestion, then it represents a useful
checksum to deduce the correct girsa. One would not expect {incorrect} "correction" of a girsa to fix this non-obvious demand for consistency, but one would more likely expect a "correction" of a text to harmonize R Yirmiya's position with the one evident in the story {that is, if one does not know of Rashi's forced explanation}. Thus, via
lectio difficilior, one could argue that our girsa is the original.
Still, we would have two slight difficulties with our sugya. The first, as mentioned above, is that Rabbi Yirmiya's statement is about a temporary attendant and in the story it seems to be a personal attendant. The second is that the attendant is not examining the tableware with an oil lamp, as in Rabbi Yirmiya's statement, but rather via a
shraga, an unspecified lamp. One can easily argue that this case was about an oil lamp, or about a lamp not of
naphta which in theory should then have the same law, as opposed to
naphta, which produces a lot more light and is unpleasant, in that one is more likely, therefore, to tip both an oil lamp or a candle, but still, the case does not exactly match his statement. We should expect the story to say that he was inspecting it via a lamp of
mishcha, oil. {
Note: the preceding paragraph was updated to note that
shraga is an unspecified lamp.}
Approaching the sugya from an academic perspective, I would suggest the following answer. From "It is no difficulty" until "It was a question to them: What of a non-permanent attendant and a [lamp fed with] oil?" is all clearly the
stama digemara. This layer is possibly
savoraic, and is post-Amoraic. We can identify it as such because it is in Aramaic, is anonymous, tries to harmonize various sources, and uses the form "And if you want I will say." Thus, I would suggest, it is a later interpolation into the gemara. If we obscure this segment of gemara for a moment, we can reconstruct the original form of the
sugya:
One brayta said: An attendant may examine glasses and plates by the light of a lamp. And another brayta said: He must not examine [them].
Rav Huna {our gemara: Rav} said: It is the halacha and we do not teach so {to others}.
Rabbi Yirmiya bar Abba said: It is the halacha and we do teach so.
Rabbi Yirmiya bar Abba visited Rav Assi. Now, his attendant arose and examined [the glasses] by a lamp {shraga}. Thereupon his [Rav Assi's] wife said to him [Rav Assi], 'But you do not act thus!' 'Let him be,' he answered her, 'he holds with his master {Rabbi Yirmiya bar Abba}.'
That is, Rav {or Rav Huna} and Rabbi Yirmiya bar Abba are dealing with the contradiction of the two
braytas. Rabbi Yirmiya bar Abba rules conclusively like the first
brayta, and says that is the
halacha, and we teach so, that an attendant may inspect tableware by the lamp. Rav says that since it is a matter of Tannaitic dispute, even though the
halacha is like the first
brayta, we do not teach so. Alternatively, he understands that the second
brayta agrees with the first one that in terms of
halacha one may, but still we do not teach so.
In general we would rule like Rav, and his harmonization of the two
braytas. However, Rabbi Yirmiya is entitled to his own opinion, and his attendant may follow that opinion. Meanwhile, no distinction whatsoever is made between a temporary or permanent attendant, or between a lamp of oil or
naphta.
After this, the
stama digemara comes and tries to harmonize the two
braytas such that they do not argue at all, and are just discussion different cases. The
stama gives one suggestion (type of attendant), and then gives a second suggestion (type of lamp) which works within one of the possibilities of the first (that is, type X attendant).
Then, some sense of and transition to the dispute of Rabbi Yirmiya bar Abba and Rav needs be made, so the dispute between the two is attributed to the case not discussed explicitly - type Y of attendant and the more stringent type of lamp.
This would account for the mismatch of the story to Rabbi Yirmiya bar Abba's statement - Rabbi Yirmiya bar Abba's statement originally was about all types of attendants and all types of lamp
1.
If we would hold like Rav {or Rav Huna}, as we seem to, based on the story, such that it is the
halacha but we do not teach so, this statement would so no matter what the attendant or type of lamp.
1) Alternatively, in the second suggestion, the stama initially discards the first suggestion, such that the the temporary or permanent nature of the attendant does not matter at all, and so says that both are attendant type X (but just as well could have been attendant type Y) but it is a difference in the type of lamp (oil vs. naphta). Afterwards, some sense of and transition to the dispute of Rabbi Yirmiya bar Abba and Rav needs be made, so the dispute between the two is attributed to the case not discussed explicitly - type Y of attendant and the more stringent type of lamp.