I’ve started reading Juan Filloy’s Caterva, one of my birthday gifts (incidentally, does anyone know the origin of the odd-looking surname Filloy?), and in the googling that inevitably accompanies my reading of such a wide-ranging book (I’ve had, for example, to read up on the Uruguayan Civil Wars) I came across the following passage from Jason A. Bartles’ ArteletrA: The Sixties in Latin America and the Politics of Going Unnoticed (Purdue University Press, 2021; free download at that link):

Karcino is a collection of Filloy’s palindromes, ranging from
two to seventeen words long, that he wrote throughout his life.
In the final section, titled “ArteletrA,” Filloy even composes
poems from his palindromes. He offers the following examples
in different languages: “Never eveN”; “Roba saboR”; “Amor
¿bromA?”; “Madam adaM”; “Bon snoB”; “Luz azuL”; and
“Amo idiomA,” among many others (74–75). Filloy prefers to
write them in capital letters to draw attention to them, since in
some cases, a seemingly simple, yet unimportant, phrase might
go unnoticed as a palindrome, such as “Acaso hubo búhos
acÁ” (81). Some can be read as poetic aphorisms, as in the case
of a seventeenth-century palindrome by John Taylor that Filloy
references: “Lewd did i live & evil i did dweL” (49). Others may
appear to be nothing more than quotidian language: “Dennis
and edna sinneD”; or “Never odd or eveN” (49). Or as in one
of Filloy’s Spanish-language palindromes: “Eufemia, jaime fue …
¡Eufemia, jaime fue!” (101).

Yet, others tend to catch one’s attention, begging to be noticed
as the ingenious constructions they are. I have selected just three

Es re-mal eros en eso: relamersE (105).
Aca, carolo adonis, amo la paloma … si no da olor a cacA
Ada, gorda drogada, di los nocivos a corola clay.
y, al calor ocaso, vi consolidada gorda drogadA (195)

A single reading will always leave the palindrome unnoticed as
such, but the capitalized letters or the comedic strangeness of
these expressions is capable of provoking a reader into giving them
a second or third glance. By reorienting my reading practices and
moving in reverse, from a different threshold of perception, the
palindrome can be perceived as such.

The Sator Square is a well-known enigma of Western cultures.
Filloy describes it as “uno de los jeux d’esprit más intrigantes de
todos los idiomas” (Karcino 59). The following is the Sator Square
as reproduced in Filloy’s treatise:


The Sator Square is composed of four Latin words, sator, tenet,
and rotas, and the unknown word arepo, which is assumed
to be a proper name but has no known, fixed meaning. As a
multidimensional palindrome, the letters in this arrangement
form a crystalline, closed structure when read from top to bottom
and from left to right and then in reverse. Its earliest inscriptions
have been dated to the first century C.E., and it has been found
among Roman ruins and in Pompeii before the arrival there of
Christianity. The Sator Square has been an object of historical and
theological speculation for centuries, since it also forms an ana-
gram for the phrase Pater noster that can be spelled twice crossing
at the letter “n”; this formation leaves as its remainder two As and
two Os, which are often interpreted as the alpha and the omega,
the beginning and the end. However, its appearance in Pompeii
challenges this possible Christian solution, and other partial
interpretations attribute it alternately to pre-Christian, gnostic,
Jewish, stoic, and even Satanic traditions. Of course, not all of
them can be correct.

Magical, miraculous, and metaphysical qualities aside,
what is certain about this and other palindromes is that they
challenge the reader’s hermeneutical skills. As Rose Mary Sheldon
demonstrates, there have been innumerable attempts at decipher-
ing this potential cryptogram since the late nineteenth century
by mathematicians, philologists, and theologians, but no one has
yet to propose a widely accepted solution to the hidden meaning
they all assume it must contain (233–87). What I find curious is
how different intellectuals can be so skeptical regarding scholarly
interpretations of the Sator Square made by others, yet these same
scholars uphold the generalized belief that this is a puzzle with a
hidden solution that has yet to be deciphered, despite all of the
contradictions present in each of these “solutions.” In the end, this
may be nothing more than a clever linguistic game centered on a
meaningless word, arepo. It may be an incidental enigma with no
hidden meaning, from which so many interpreters extract only
what they wanted it to contain a priori.

I like that last sentence very much; it expresses concisely and forcefully something I often feel when reading far-fetched analyses. We’ve discussed palindromes before, e.g. in 2011 and 2023; oddly, we seem never to have mentioned the Sator Square.


  1. Dmitry Pruss says

    the oldest examples of Filloy in the vital records indexed by familysearch are mostly in France (but one in Switzerland), and there, mostly in Eure-et-Loir in the 1500s

  2. Long palindromes must be easier in some languages. Lack of consonant clusters; Few inflection; Absent or ignored diacritics; ,,,

  3. Absent or ignored diacritics

    In Hebrew for sure. Hebrew palindromes thus don’t usually suffer from the implausible syntactic violence which English ones so often require. יֶלֶד כּוֹתֵב בְּתוֹךְ דְּלִי ‘a boy is writing inside a bucket’ is silly but grammatically unremarkable. Some palindromic verse is actually good beyond novelty value.

  4. There’s a famous Hebrew SATOR square. The story goes, Abraham Ibn Ezra was asked if a fly or a bee which have drowned in honey make it impure. His answer was,
    That is, ‘We interpret: the glutton in the honey was incinerated and was burned’: the offending insect was dissolved, and the honey is fit to eat.

    The earliest reference to the palindrome and to the story which I could find is here (Jüdisches Volksblatt 1(26), 104, 1854), as an “anecdote” told by “L. B. in D.” I haven’t found any proven evidence of its actual authorship.

    (Can you do fixed-width font here?)

  5. The first word, פרשנו, means ‘we interpreted’.

  6. Owlmirror says

    Does [code] work?


  7. fixed-width font

    The sofit kind of spoils it, too, doesn’t it?

  8. The sofit kind of spoils it, too, doesn’t it?

    A very reasonable assumption, but not really. You don’t notice it much. Just like the reverse — in crossword puzzles, you don’t use the final forms, and it registers as only slightly off.

  9. Filloy is a Galician (not Galitzian!) surname. Spanish Wikipedia says that he was “hijo de Benito Filloy, un campesino español de Pontevedra, Galicia, y de Dominique Grange, una campesina francesa natural de Toulouse, que se ganaba la vida como curandera y lavandera”
    There are currently about 400 Filloys living in Spain, mostly in Galicia.

  10. A toponymic surname? Filloi is also the name of two places in Galicia. Some speculation on the origin of their name here (scroll down):

    4) Filloi (67 hab.- Forcarei, Pontevedra)
    Filloi (24 hab.- Sarria, Lugo)
    Parecen, ambos, genitivos de antropónimos. De hecho, las formas inmediatas más probables serían *Filioli, *Filioni o *Filiodi. Ninguna de ellas se corresponde con un antropónimo que yo conozca, pero la primera es el genitivo del latino FILIOLUS “hijo adoptivo”, que habría dado un castellano hijuelo y un gallego filló. Sin embargo, no encuentro en el CODOLGA que se usara como nombre propio en Galicia. De hecho, el único antropónimo documentado en la Galiciamedieval que conozco, y que fonéticamente pueda haber generado los anteriores topónimos, es Villoi/Uilioy (Sobrado 818, Lalín 956), derivado del tema germánico *weljan “Voluntad”.
    No conozco ninguna etimología más inmediata, pero agradezco aportaciones.

  11. The sofit kind of spoils it

    Aww a truss and soffit would balance it nicely.

    If I’m understanding reports correctly, a Truss has become available unexpectedly.

  12. Abbas and Xerîb: Many thanks! It seems an actual etymology can’t be had (although I do like the FILIOLUS speculation), but those are the breaks, and I now know as much as can be known.

  13. Long palindromes must be easier in some languages

    Finnish is definitely easy mode for palindromes (vowel harmony, unstressed long vowels possible, no initial consonant clusters, many reversible medial consonant clusters and diphthongs / vowel clusters) — Finnish Wikipedia reports our longest composed palindromic poem to be 170,000 letters… and for general illustration, here is one single-sentence classic I’ve memorized from the humorist group Alivaltiosihteeri, who’ve published several books worth of these:

    Taka–Loimaan Tekun taloustillipaalit ajat lavarekalla, Matti; osat saat heti tehtaasta soittamalla Keravalta; ja tilaa pillit, suolat, nuket, naamiolakat!

    “The household dill bales of Teku in hinterland Loimaa [town northeast of Turku] you will drive with the pick-up truck, Matti; the parts you will receive immediately from the factory by calling from Kerava [town north of Helsinki]; and order the straws, salts, dolls, mask lacquers!”

    The nonce name “Teku” and its corresponding nuket can be even cleanly removed, if we feel these are inelegant in palindromes.

    and I designed a Finnish Sator square myself some ten years ago:


    “Plucking [‘raising’]; sprouts; the hundredth taro [is] the bear’s”

  14. If I’m understanding reports correctly, a Truss has become available unexpectedly.
    Buyers beware, seems to be bad quality, short-lived material.

  15. David Marjanović says

    Amor ¿bromA?

    There’s a much better one in Pompeii.


  16. Acaso hubo búhos acÁ

    Why the accent in búho? Isn’t that standard penultimate stress?

  17. Because otherwise it could be pronounced as one syllable. As lazarus1907 says:

    Vowels can be strong (A, E, O) or weak (I, U). Two strong vowels together, or two weak vowels together belong to different syllables: a-o, a-e, o-a,….The combination of a weak vowel and another vowel gives a diphthong: ai, au, ei, ia, iu,…”buho” without the accent would be one syllable (I’ll ignore the h): “buo”, and the accent would be on the “o”. This would be a diphthong. If you want to stress the “u”, you have to break the diphthong: “bú-ho”.

    Or more concisely here:

    Se tilda para romper el diptongo /uo/: si no se tildara sería una palabra monosílaba /buó/; gracias a la tilde es bisílaba /bú-o/.

  18. Rodger C says

    I learned buho in 1962. When a later spelling reform introduced the agudo, I was like, wtf? AFAICT, the h (a consonant realized as zero) makes the pronunciation unambiguous without it. If the Real Academia wants the h to be utterly ignored, why don’t they remove it from the alphabet?

  19. Lars Mathiesen (he/him/his) says

    The RA wants (intervocalic) h to be utterly ignored for the purposes of syllabification and accentuation. I don’t think you’re allowed to not write it in cultismos like adhesivo. Also I don’t understand its role in forms like huele from oler, maybe people in the 16th or so just didn’t like the look of uele. (Or uole, maybe, I don’t know when the diphthong changed. Also things like huérfano, I think it’s pretty consistent).

  20. David Eddyshaw says

    a Truss has become available unexpectedly

    When I lived in Bawku, one of the (non-Kusaasi) drivers had the surname “Trust.”

    In accordance with the usual construction of Kusaal personal names, almost all of which take the initial particle A-, he was always actually called Atrɔs. In fact, until I happened one day to see his name written down, I always thought he was actually called “Atroce” (à la française.) He did have a somewhat equivocal reputation locally, but it did seem a bit pointed, somehow.

    I don’t know why that came to mind just now.

  21. Keith Ivey says

    two weak vowels together belong to different syllables … The combination of a weak vowel and another vowel gives a diphthong

    Those statements are contradictory. Only the second is true. Two weak vowels (iu or ui) make a diphthong, as in ciudad or cuidado.

  22. Thanks for the explanation, hat.

    Not that anyone should care about my opinion, but I’m with Rodger C. If plain uho is supposed to be pronounced like uo, then the h is worse than useless.

  23. Lars Mathiesen (he/him/his) says

    So how do you tell from the written form if two high (“weak”) vowels without accents form a rising or a falling diphthong? Gut feeling? IIRC, the RA is silent on the matter. There’s fuimos (rising) besides muy (falling). (But the falling ones are rare, I couldn’t quickly find one that wasn’t word-final and spelled with y. Even a word like arruinar seems to have a rising diphthong, where I would find it easier to use a falling one. Also no example in of falling iu).

    The current dissatisfaction about spelling rules is about the (recent) decision that forms like vio are monosyllables (rising diphthongs) and unlike forms like salió they do not take a tilde. The unspoken assumption is that high-low always diphhongizes in the modern standard, but there’s a concession to very conservative speakers who may actually have hiatus / two syllables in such words and are allowed to write vió, guión if that’s how they say it.

    (There’s actually no way to mark in writing if you have a three-syllable salió with hiatus before the ó, but that’s not new).

  24. @Rodger C:

    If the Real Academia wants the h to be utterly ignored, why don’t they remove it from the alphabet?

    Because that would mean acknowledging that reformers like Bello had it right even though they were from the Americas.

  25. Or uole, maybe, I don’t know when the diphthong changed

    Was there ever an [uo] in Ibero-Romance? I recall seeing an explanation that the process of Romance open-mid vowel diphthongization was rather along a track of *ɔ > [ɔə] > [uə] > [ue], [uø], [uo] etc.

  26. Lars Mathiesen (he/him/his) says

    I got my info from this, which is a pretty solid site AFAICT. (The lecturer maintaining it does publish on the historical phonology of Spanish, but that specific section is rather cursory. It does have the point that it’s /uo/ in Italian, but that’s not probative: Iberian romance could have skipped that step).

    (The site is old enough to use frames so you can’t navigate to the front matter from the link I gave, and the subsection link [#fragment] doesn’t propagate through the links that do show the frame).

    I’m sure there are people who have a more detailed account, I never saw it though. Mackenzie does cite Menéndez Pidal, which I am reminded I should find some day and read when I have time off, but that was published 37 years ago.

  27. Rodger C says

    Ah yes, the good old Irreal Academia.

  28. David Marjanović says

    Was there ever an [uo] in Ibero-Romance? I recall seeing an explanation that the process of Romance open-mid vowel diphthongization was rather along a track of *ɔ > [ɔə] > [uə] > [ue], [uø], [uo] etc.

    Indeed, in High German, the order in which the spellings are attested is AFAIK ea, then ia and then ie, likewise oa, then ua and then ue. I don’t know if that’s actually Romance spillover – the fact that oo spellings occur in the earliest Bavarian sources and are only replaced later may argue against that; but in any case this looks like it went [ɛː] > [ɛɐ] > [ɪɐ] > [ɪɛ] or suchlike, with no [e(ː)] ever involved, and likewise for [ɔː] (remarkably enough).

    They’re [ɪɐ̯ ʊɐ̯] in my dialect, though that may be connected to the facts that 1) [ɐ] is the (rare) reduced vowel and 2) non-rhoticity has turned [ɪr ʊr] into [ɪɐ̯ ʊɐ̯] anyway.

  29. David Marjanović says

    I just came across a citation of the etymological dictionary of OHG by Braune & Reiffenstein (2004) accompanying the statement that “*ē₂” was still “ē” in the 8th century, “soon” ea showed up alongside it, and in the 9th century it turned into ia & ie.

Speak Your Mind