Here's a coupla poems for ya — specifically, #627 and #628 from the Man'yōshū:

吾手本 将巻跡念牟 大夫者 戀水定 白髪生二有
Wa ga tamoto/ makamu to omopamu/ masurawo pa/ namita ni sidumi/ siraka opwinitari
That worthy would rest his head on my arm; sinking into tears, my hair has begun going gray

白髪生流 事者不念 戀水者 鹿煮藻闕二毛 求而将行
Sira ka opuru/ koto pa omopazu/ namita wo pa/ ka ni mo kaku ni mo/ motomete yukamu
I didn't even notice the gray hairs; I will at all costs go and seek these tears

(Here and throughout I am mixing-and-matching my text from Satake et al 1999, Frellesvig et al 2014, Ide and Mōri 2009, and two editions from 1643 and 1709 Waseda University has put online; see in particular here and here. Romanized orthography is based on Frellesvig et al 2014.)

Oh, wait, that's just how folks read it for hundreds of years. Here's what modern textual scholarship has to say:

吾手本 将巻跡念牟 大夫者 變水求 白髪生二有
Wa ga tamoto/ makamu to omopamu/ masurawo pa/ wotimidu motome/ siraka opwinitari
That worthy who would rest his head on my arm, let him seek the water of youth: his hair has begun going gray

白髪生流 事者不念 變水者 鹿煮藻闕二毛 求而将行
Sira ka opuru/ koto pa omopazu/ wotimidu pa/ ka ni mo kaku ni mo/ motomete yukamu
I didn't even notice the gray hairs; I will at all costs go and seek this water of youth

Here's what's going on.

The third line of the old version includes the characters 戀水, literally "love-water", here to be read /namita/ "tears." This is found in a large number of manuscripts, including the late Kamakura Nishi-Honganji-bon 西本願寺本. As the NHB is the oldest complete manuscript, its authority has long been respected. I can't think of a modern edition that doesn't use it as a base text, except for some exceptional cases like John Kupchik's edition of the Eastern Old Japanese corpus, which he based on the Heian Genryaku kōhon 元暦校本 because, he says, it contains "a variety of EOJ features that are not in NHB or later manuscripts" (Kupchik 2011: 910).

As it happens, the Genryaku kōhon ("[critical] edition") is key to these two poems too. The GK is dated to 1184 (Genryaku 1, hence the traditional name), and is the third oldest MYS manuscript of all, containing just over half the poems (Vovin 2009: 11).

The GK has 變水 ("change-water") instead of 戀水 ("love-water") for poem #627, which opens just wide enough a gap for modern scholars to see the OJ word /wotimidu/ inside. /wotimidu/ comes from /wot-/, a verb meaning "return [to the beginning, to one's youth]" and /midu/, ancestor of the contemporary mizu, "water". Elsewhere in the MYS it is said to be in the possession of Tukuyomi, the moon deity. (All that waxing and waning, you know.)

Meanwhile, the character that is variously interpreted as "sink", "seek" (and in other traditions, /sadamu/ "stabilize" — "pacify" for the second poem) is 定 in the old manuscripts, but this was apparently reinterpreted as 求 under a 宀 by Takagi, Gomi and Ōno in their 1957-1962 edition of the MYS (Ide and Mōri 2009: 97; Satake et al 1999: 379), and from there it's easy to get to 求, "seek". Which you want to do, because the second poem has 求 in all manuscripts as far as I can tell.

So what? The main thing that interested me about all this is that the original and (let's stipulate) correct reading of these poems seems pretty weak. I mean, look at it: "If you want to be my lover, find the fountain of youth, because your hair is going gray" → "Going gray, you say! I didn't notice! Anyway, I'll get on that fountain of youth thing right away." Where is the conflict, the repartee... the wit?

The traditional version may be forced (whoever annotated the 1709 edition linked above used a lot of red ink trying to explicate these two poems) and, okay, incoherent, but if you allow /sadamu/ with a meaning like "soothe" in the second you get a progression that at least seems to be going somewhere: "Sleeping with you will be a vale of tears and make my hair go gray" → "That won't bother me; let me dry those tears."

There's also the use of "love-water" to mean "tears" — that's so exactly the sort of thing you expect to find in the Japanese poetic corpus (albeit more Heian than Nara) that it almost seems like an intentionally laid trap for those too confident in their own understanding of the tradition. (On the other hand, the alternative — a reference to magical enyoungening water — is so pleasingly idiosyncratic that it could be a trap for lectio difficilior potior zealots.)

Works cited

• Frellesvig, Bjarke; Horn, Stephen Wright; Russell, Kerri L; and Sells, Peter. The Oxford Corpus of Old Japanese. 2014. <http://vsarpj.orinst.ox.ac.uk/corpus/>.
• Ide, Itaru 井手至 and Mōri, Masamori 毛利正守, eds. Shin Kochu Man'yōshū 新校注万葉集. Tokyo: Izumi Shoin, 2008,
• Kupchik, John. "A Grammar of the Eastern Old Japanese Dialects." Diss. University of Hawaiʻi, 2011.
• Man'yōshū. Seiganji-mae, Kyoto: Yasuda Jūbei 安田十兵衛, 1643. <http://www.wul.waseda.ac.jp/kotenseki/html/bunko30/bunko30_d0001/index.html>.
• Man'yōshū. Izumoji Izuminojō 出雲寺和泉掾, 1709. <http://www.wul.waseda.ac.jp/kotenseki/html/he04/he04_00090/index.html>.
• Satake, Akihiro 佐竹昭広; Yamada, Hideo 山田英雄; Kudō, Rikio 工藤力男; Ōtani, Masao 大谷雅夫; and Yamazaki, Yoshiyuki 山崎福之, eds. Shin Nihon Koten Bungaku Taikei 1: Man'yōshū 1 新日本古典文学大系一 万葉集一. Tokyo: Iwanami Shoten, 1999.
• Vovin, Alexander, ed. and trans. Man'yōshū: Book 15. Folkestone, Kent: Global Oriental, 2009.