κνέφαλ(λ)ον, τύλη
(Phryn. Ecl. 145, Moer. κ 42, Poll. 10.39–41, Philemo [Vindob.] 393)

D. General commentary

Atticist scholarship discusses the word κνέφαλ(λ)ον (‘wool flock’, hence ‘cushion’) in two respects: spelling and word choice. Regarding the spellingSpelling, κνέφαλλον is addressed within the context of a broader discussion of the verb κνάπτωκνάπτω    (‘to full, to comb’) and its derivatives, for which Atticist sources adopt and promote the Attic spelling with the cluster κν- against the more innovative voiced spelling with γν- (on this issue, which will not be discussed in what follows, see the entry ἄκναπτος and AGP vol. 2, Phonology, forthcoming). Regarding word choice, Atticist scholars consider κνέφαλλον to be the approved Attic word for ‘cushion’ and endorse its use over τύλη (A.1, A.2, A.4). The subsequent paragraphs will first provide a brief analysis of the history of these words and then investigate the reasons behind the Atticists’ terminological preference.

As we have seen, the word κνέφαλλον belongs to the group of derivativesDerivatives of κνάπτω and properly denotes the wool flock torn off during the fulling process. Hence, κνέφαλλον is also used by synecdoche to signify not only the stuffing material, but also the stuffed object itself (as mentioned in B.1). Although this synecdochical meaning is classical and typical of Attic (see below), some scholars possibly rejected it, as Pollux records (A.3). Despite the challenges in precisely classifying the ancient vocabulary related to furnishings and textiles, Rotroff (1978, 198) is most likely correct in identifying the κνέφαλλον as ‘the large cushion or mattress on the κλίνη’. Originally, at least, the name must have distinguished the object’s material, and κνέφαλλον was possibly limited to textiles padded with wool (unlike, for instance, the προσκεφάλαιον, a pillow specifically designed for resting the head and made of any cloth, see Andrianou, Miramontes, Quillien 2024, 172). κνέφαλλον (in the Aeolic form γνόφαλλον) is first attested in Alcaeus (fr. 338 Voigt), where it refers to a woollen garment worn around the temples for protection against the cold, probably a sash. It is then found in Euripides (fr. 676, meaning cushion or mattress), and has multiple occurrences in Attic comedy: besides C.3, C.4, and C.5, see also Eup. fr. 218. In C.4 and C.5, κνέφαλλον is certainly used in its primary sense to denote wool-flocks used as stuffing material (in C.4 it is used to mock the pomposity of a soldier who is so bombastic that he appears to be stuffed with wool-flocks or feathers, see Pirrotta 2009, 206; on C.5 see Farmer 2022, 146–9). When it comes to the synecdochical meaning found in C.3, it is difficult to determine precisely what object that κνέφαλλον denotes. The κνέφαλλον requested, alongside a προσκεφάλαιον, from a bedroom could indeed refer to either a mattress and a cushion. In the first case, the text would imply a pair consisting of a mattress and a pillow; in the second, it could refer to two types of cushions, perhaps made of different materials (wool and linen, respectively). Although definitive evidence is lacking, the impression is that κνέφαλλον here denotes a mattress; this is the interpretation adopted by Orth (2017, 114), who, while leaving both possibilities open (116–9), reasonably chooses ‘Matratze’ in the passage’s translation.

Epigraphical occurrences of the synecdochical use of κνέφαλλον provide important evidence of its specifically Attic usage. κνέφαλλον meaning ‘mattress’ notably occurs in the Attic Stelai, as Pollux recalls (A.3, see Pritchett 1956, 247–8; on the relevance of inscriptions for Pollux, see also Zadorojnyi 2018); in addition to C.8, κνέφαλλον is also attested in the contemporary IG 1³.422.col.iii.259–60; in both lists the object at issue is understood as a mattress. A κνέφαλλον is also mentioned in C.9: here this object’s position between a bed (κλίνη) and a bed covering (περίστρωμα) suggests that it, too, refers to a mattress (cf. Rotroff 1978, 198; see also Orth 2017, 119).

The history of κνέφαλλον does not end with classical Athens, although its usage appears to have declined in later ages. In the imperial age, κνέφαλλον (often in the voiced spelling γνάφαλ(λ)ον: see above and the entry ἄκναπτος) is used, among others, by Strabo (15.1.20.24), Dioscorides (117.1), Soranus (2.16.2), and Galen (e.g. De simpl. med. fac. 11.161.4–5 Kühn); for a Lucianic occurrence discussing the word’s spelling, see the entry ἄκναπτος. Remarkably, κνέφαλλον does not denote a textile of any kind in any of the above-mentioned passages, but simply the wool-flock: the synecdochical meaning that was widespread in classical Attic appears to have fallen out of use in later ages, except in erudite discussions. This is clearly illustrated in C.10, where the mattress (or, less likely, the cushion) itself is referred to as τύλη, while the wool-flocks with which it is stuffed are called γνάφαλλα. The fact that κνέφαλλον was not particularly popular after classical times is seemingly confirmed by its absence from extant papyrological documentation.

As for τύλη, the word encompasses a range of meanings unified by the notion of ‘swelling’: in addition to referring to a padded textile (again, either a cushion or a mattress: see below), it can also denote, among other things, a swelling of the skin, such as a blister or callus (cf. e.g. Ar. Ach. 860, this meaning is recorded in Hsch. γ 831 and Phot. τ 541, tentatively identified with Ael.Dion. τ 28; see also schol. Ar. Ach. 954b). The etymology of τύλη is uncertain (for an overview of tentatively proposed cognates, see DELG s.v.; EDG s.v.). Chantraine (cf. DELG s.v. τύλη, τύφη) proposes a connection with τύμβος (‘tomb’, originally referring to a burial mound) and, ultimately, with τύφη (Typha angustifolia, a plant used for stuffing cushions and mattresses, according to Theophrastus, e.g. HP 1.5.3 – see LSJ s.v.), all deriving from an IE root *tuH-. τύλη has a masculine equivalent, τύλοςτύλος, which can denote a callus or any protruding object (including the penis, cf. Poll. 2.176Poll. 2.176), as well as, more rarely, a cushion or mattress. The diminutives τυλεῖον and τυλάριον also exist. While the first form, τυλεῖον, is attested as early as Sophocles and is presented as Attic by B.2 (see below), τυλάριον is most likely a later formation: its only known occurrence in the BCE period is in P.Genova 2.56 A,2.1 (= TM 78853) [Arsinoites, middle of the 3rd century BCE]; it later occurs in documentary papyri and literary sources from the imperial and Byzantine periods (cf. LBG s.v. τυλάριν). τύλη, τύλος, τυλεῖον, and τυλάριον all denote a padded textile in general, which, similarly to κνέφαλλον, must be identified on a case-by-case basis (and not without some difficulty) as either a cushion or a mattress (see Clarysse, Geens 2009, 40–1; Borrelli 2017, 120–1; Andrianou, Miramontes, Quillien 2024, 176; Wipzycka 1965, 119–20, instead, only mentions the meaning ‘cushion’).

τύλη is attested as early as Sappho (C.1, on which see Neri 2021, 648–9) and, despite Herodian’s claim in B.1 that it is not used by Attic authors (οὐκ ἦν παρ’ Ἀττικοῖς), it is not without attestations in Attic – even in the specific meaning at issue (see its occurrences in C.6 and C.7, as well as the diminutive τυλεῖον in C.2). As we have seen, it is difficult to determine with certainty what object τύλη refers to, and translations often diverge. In C.1, both a soft mattress (cf. Neri 2021, 432: ‘coltrice’; 648: ‘materasso’) and a soft cushion (cf. Campbell 1990, 93: ‘cushions’) would satisfy the desire of the persona loquens. In C.7, τύλη occurs in a list of items related to bedding – perhaps in preparation for a banquet (see Olson 2021, 90–1) – and may denote either a cushion or a mattress, in the latter case functioning as a quasi-synonym of the στρῶματα of the preceding verse. In C.6, further context would be required to identify the specific τύλη at issue. Scholars have made different suggestions regarding the meaning of τύλη in this passage, largely in the light of Pollux’s claim (A.3) that Eupolis ‘speaks Ionic’ (ἰάζοντι) when using this term. This statement, however, lacks any linguistic basis – since the word is not Ionic nor treated as such by any other erudite source – and may instead be explained in view of semantic issues. Cassio (1981) suggests that the passage may preserve a mocking allusion to Protagoras’ Ionic softness, conveyed through the mention of a soft, stuffed support (τύλη) and hairnets (κεκρύφαλοι) – a typically feminine accessory often used to ridicule male effeminacy (cf. Ar. Th. 138, 257). According to Corradi (2005) and Napolitano (2012), instead, τύλη here specifically alludes to the anecdote that Protagoras invented a porter’s pad to ease the carrying of burdens – an object referred to, precisely, as τύλη (see LSJ s.v. 2; for the anecdote, cf. D.L. 9.53 = Protag. Diels–Kranz 80 A 27–8 = Arist. fr. 72 Gigon). According to this reconstruction, C.6 depicts the intellectual Protagoras as poised between refinement and concreteness (Corradi 2005, 404; Napolitano 2012, 123). On Eupolis’ fragment, see also Olson (2016, 78–80).

τύλη is the standard form in both the high and low koine, as evidenced by ancient scholarship and the frequent number of occurrences of this word in documentary papyri. Moeris (A.2) assigns τύλη to the Ἕλληνες, a label here likely referring to the literary koine of his time (see entry Moeris, Ἀττικιστής, with bibliography). Pollux (A.3) similarly states that Antiphanes (C.7) adheres to the ‘common usage’ (κατὰ τὴν κοινὴν χρῆσιν) in using τύλη in place of the specifically Attic κνέφαλλον (on the notion of κοινὴ χρῆσιςχρῆσις in Pollux, with particular reference to this passage, see Valente 2013, 147–8 and below). When it comes to documentary attestations, a search on Trismegistos shows that τύλη occurs in 35 papyri (in some of them multiple times, for a total of 42 occurrences), spanning from the 3rd century BCE and the 4th CE. Α good example of the use of τύλη in everyday language is C.11 (Philadelphia, Arsinoites, 59 CE), a private letter addressed by a woman, Thermuthis, to Nemesion, possibly her husband. Here a τύλη, presumably a mattress, is mentioned among other textiles and furnishings to be exchanged with a certain Lucius (see further Bagnall, Cribiore 2006, 72, on Nemesion; and 344 on the letter’s content and form).

The success of τύλη – an ancient word, as indicated by its likely IE root and its occurrence in Sappho (C.1) – at the expense of κνέφαλλον in the koine may be explained, following Pearson (1917 vol. 2, 120), as follows. Originally, τύλη and κνέφαλλον denoted distinct objects: respectively, the stuffed textile (either a mattress or a cushion) and the wool stuffing. The synecdochical use of κνέφαλλον for the textile object then caught on in Attic and became a peculiar aspect of Attic usage (cf. B.1, where the use of the same word for both the stuffing and the stuffed object is presented as peculiar to Attic authors). Later, the koine abandoned this specifically Attic usage and retained the standard one, in which τύλη served as the normal term for cushions and mattresses.

In recommending the use of κνέφαλλον over τύλη, Phrynichus (A.1) thus offers an anachronistic prescription that can, nevertheless, be readily explained in the light of the preceding analysis of these two words’ history and diastratic distribution. Although Phrynichus acknowledges that τύλη is also used by Attic authors – albeit with only a rather vague reference (εἰ καὶ εὕροις που) – a strict Atticist like him could not help but prefer κνέφαλλον. As we have seen, the synecdochical use of κνέφαλλον is a peculiar feature of Attic usage and would have been perceived as a mark of finesse compared to τύλη, which had not only become dominant in koine Greek, but was also common in its lower registers. The same uncompromising stance is adopted by Moeris (A.2), Philemon (A.4), and Herodian (B.1), with the latter (wrongly) maintaining that τύλη never occurs in Attic authors. Pollux’s position (A.3) is instead less intransigent, and his mention of the ‘common usage’ should not be understood as disapproval: in noting that a word belonging to the κοινὴ χρῆσις like τύλη also finds room in ancient authors, he allows for the possibility of using them (on Pollux’s intention to offer an exhaustive picture of the ancients’ usus as a benchmark for contemporary vocabulary, see Valente 2013; note also that, according to the now outdated but formerly popular reconstruction proposed by Naechster 1908, 27, this is one of the passages in Book 10 that Pollux wrote in response to Phrynichus’ criticism). It is impossible to determine Aelius Dionysius’ stance on this issue, given that he is the source of B.4. While this identification is not implausible, it lacks proof. Later erudite sources, perhaps resting on a simplified version of the Atticist materials, mostly confine themselves to explain κνέφαλλον and τύλη as synonymsSynonyms (B.2, B.3, B.4, B.5, B.6), sometimes providing additional information on the spelling (B.3) of the former. It is noteworthy that B.2 and B.5 suggest that κνέφαλλον and τύλη are synonymous with προσκεφάλαιον. One may wonder whether the association of κνέφαλλον and προσκεφάλαιον mirrors their pairing in Aristophanes’ fragment (C.3, κνέφαλλον ἅμα καὶ προσκεφάλαιον) and, similarly, whether the mention of a πτιλόν in B.2 reflects C.4 (κνεφάλλων ἢ πτίλων). It is indeed plausible, though not provable, that B.2 is the remnant of an ancient, more extensive discussion encompassing the passages from Aristophanes and Plato. It is also noteworthy that B.2 belongs to a group of items in Hesychius’ lexicon, the so-called Attic interpolations, which were added at a later stage. In B.2, as in other items of this group, the pronoun ἡμεῖςἡμεῖς is used as an identity label marking a koine form that is negatively connotated in Atticist scholarship (see Matthaios 2010, 92–3, and AGP vol. 3, forthcoming). Regarding Thomas Magister (B.6), the last scholar to discuss these forms, it is remarkable that he does not align with the intransigent Atticists (see A.1, A.2, A.4), as would be expected. Instead, after rejecting the voiced spelling of κνέφαλλον, he presents its equivalence to τύλη in a non-proscriptive manner (note that this item is already present in cod. Ferrara gr. II 155, f. 155r, the redaction that most closely mirrors Thomas’ original work; on the Eclogue’s transmission, see Gaul 2007 and entry Thomas Magister, ’Ονομάτων Ἀττικῶν ἐκλογή).

When it comes to the diminutive τυλεῖον, C.2 is its only attestation, except in erudite sources. The notion that it is Attic in B.2 most likely depends on knowledge of the Sophocles passage. Indeed, like τύλη, τυλεῖον must also have been in use in the koine, as its occurrences in documentary papyri suggest: cf. O.Deiss. 62.7 (= TM 73877) [Apollinopolis, 187 or 163 BCE], P.Oxy. 14.1645.9 (= TM 21954) [308 CE], and P.Vindob. Worp 24.4 (= TM 30404) [unknown provenance, after 324 CE].

E. Byzantine and Modern Greek commentary

Both κνέφαλλον and τύλη are attested in Byzantine prose, albeit infrequently. As for κνέφαλλον, the Atticist prescription of the voiceless spelling    appears to have had little impact, since outside of erudite sources it is mostly spelled γνάφαλ(λ)ον. The word occurs, for instance, in Paulus of Aegina’s Medical Compendium in Seven Books (7.3.3.90) and in the Geoponica (10.21.10, 10.28.2). Occurrences of τύλη are instead found, for instance, in Pseudo-Mauricius’ Strategicon (10.3.2), Constantine VII Porphyrogenitus (De virtutibus et vitiis 1.234.2), and Georgius Cedrenus (Compendium historiarum 2.377.13). The neuter τύλιον (‘Matratze’ according to LBG s.v.) is given as an equivalent of τύλη by Theognostus in his Canones (135.3) and by [Zonar.] 1755.12. The diminutive τυλάριον (on which see above) is also sometimes used.

Neither of these words is still in use in Modern Greek, where the standard words for ‘mattress’ and ‘cushion’ are the already ancient στρώμα and μαξιλάρι, respectively. The latter entered Medieval Greek from Latin (cf. Lat. maxilla ‘jaw’, maxillaris ‘of the jaw’; see Dickey 2023, 282) and is attested, in the form μαξιλλάριον, as early as the 9th century.

F. Commentary on individual texts and occurrences

N/A