Book Nav

22. The perfect stem is formed by reduplication, and is liable to vary with the person endings (§ 6). This variation is the rule in the Homeric perfect. In Attic it survives in a few forms only; it is regular in οἶδα and ἕστηκα. The weak form of the stem is the same (except for the reduplication) as in the tenses already discussed. The long stem is often different, showing a predilection for the ο-form. The variation appears in the interchange of

  1. -η- (-ᾱ-) and -ᾰ-
     

    • τεθήλ-ει
      part. fem. τεθᾰλ-υῖα
      bloomed
       
    • ἄρηρε
      part. fem. ἀρᾰρ-υῖα[/fn]
      is fitting
       
    • λεληκ-ώς
      part. fem. λελᾰκ-υῖα
      yelling
       
    • μεμηκ-ώς
      part. fem. μεμᾰκ-υῖα
      bleating
       
    • λέλασται
      (λελαθ-ται, λήθ-ω)
      has forgotten
       
    • ἀκαχ-μένος
      sharpened
       
       
    • πέφαν-ται
      has appeared
       
       
    • σέσηπε
      (σαπρός)
      is rotten
       
    • τέτηκα
      (τήκ-ω)
       
    • τέθηπα
      aor. part. ταφ-ών
       
    • πέπηγε
      (πάγ-η)
       
    • κεχην-ότα
       
       
    • κεκληγ-ώς
       
    • πεπληγ-ώς
       
    • τετρήχ-ει
      (τᾰρᾰχ-)
       
       
    • πεπάσ-μην
      (πατ-έομαι)
      I had eaten
       
    • κεκασμένος
      (κᾰδ-)
      excelling
       
    • ἐρράδ-αται
      are sprinkled
       
       
    • δέδασ-ται
      is divided

    but 3rd plur. δεδαίαται, from δαι-, § 51.2).

    In the last four cases the strong form does not actually occur.

    δέδηε is on fire is for *δέδηυε (δεδηϝ-ε): the weak stem is δᾰυ- (δαίω for δαϝ-ι̯ω, cp. καίω, ἔκηα). Similarly γέγηθε rejoices is for *γέγηυθε (Lat. gaud-eo).

    ᾱ for η occurs in ἔαγε is broken (Hes. Op. 534: ἐάγῃ as subjunctive is only Bekkerʼs conjecture in Il. 11.558, see § 67): also in ἑᾱδ-ότα pleasing, as to which see § 26.2.

    ω and ᾰ

    This interchange cannot be exemplified from Homer: cp. Attic ἔρρωγα (ῥᾰγ-, middle συν-έρρηκ-ται). -ω- is also found in ἄνωγα I bid, γέγωνε calls aloud, but the corresponding weak stems are unknown.

  2. ω and ε

    εἴωθε[fn]Cp. ἔθων, ἦθος, root σϝηθ-.[/fn]
    is accustomed

    ἐπ-ώχ-ατο (from ἐπ-έχω)
    were shut to (of gates)

    συν-οχωκ-ότε[fn]Better perhaps συν-οκωχότε, see Cobet, Misc. Crit. p. 303; from συν-έχω (cp. ὀκωχή a stay or buttress, ἀν-οκωχή = ἀνοχή staying, cessation.)[/fn]
    leaning together

    η and ε

    μέμηλε
    is a care

    ἐδ-ηδ-ώς
    having eaten

  3. ω and ο
    • δέδο-ται (δω-)
       
       
    • ἐκ-πέπο-ται
      is drunk up
       
    • ὄλωλε
      is lost
       
    • ὄρωρε
      is aroused
       
    • ὄπωπα
      have seen
       
    • ὀδώδ-ει
      smelt

    perhaps also ὀρώρει watched (Il. 23.112 ἐπὶ δʼ ἀνὴρ ἐσθλὸς ὀρώρει = was the ἐπί-ουρος), cp. § 30

    προ-βέβουλα (Il. 1.113) seems to follow the pres. βούλομαι: we expect *βέβωλα (βολ-, § 30).

  4. οι and ῐ
     

    • οἶδα
      1st plur. ἴδ-μεν
       
    • πέποιθα
      1st plur. pluperf. ἐ- πέπιθ-μεν
       
    • ἔοικα
      dual ἔϊκ-τον
      part. fem. ἐϊκ-υῖα
       
    • λέλοιπα
      aor. ἔ-λῐπ-ον
       
       
    • δείδω[fn]for δέδϝοια (by loss of -ι- and contraction)[/fn]
      1st plur. δείδι-μεν
      (for δέδϝῐ-μεν)
      fear

    This account of the isolated 1st singular δείδω was given by G. Mahlow (K. Z. xxiv. 295), and has been adopted by most scholars. The original Homeric form was probably δείδοα (or δέδϝοα), which can be restored in all the passages where the word occurs. Others (as Cobet) would substitute δείδια, a form which is found in several places, sometimes as an ancient v. Il. for δείδω, But it is difficult on his view to account for the change from δείδια. Rather, an original δείδοα (or δέδϝοα) was altered in two ways, (1) by contraction, which gave it the appearance of a present in -ω, and (2) by change of ο to ῐ under the influence of δείδῐ-μεν, etc.

  5. ευ and ῠ

    πεφευγ-ώς
    mid. πεφυγ-μένος
    having escaped

    τετεύχ-αται
    3rd sing. τέτυκ-ται
    are made

    κέκευθε
    aor. κῠ́θε
    hides

    ἐζευγ-μένοι
    (ζυγ-όν)
    joined

    Other weak stems: κέχῠ-ται, ἔσσῠ- ται (§ 15), πέπυσ-μαι (πῠθ-), κέκλῠ-θι listen.

    ου interchanging with υ is much less common: εἰλήλουθα I am come (ἐλῠθ-), perhaps δεδουπ-ότος (cp. κτύπ-ος).

    ῡ appears in μέμῡκε (aor. μῠ́κε), βέβρῡχεν roars, as in the present μυκάομαι, βρύχω.

  6. ορ (ρο), ολ and ᾰρ (ρᾰ), ᾰλ (for , , § 6.5)
     

    • δι-ἐφθορας
      (φθᾰρ-)
      are destroyed
       
    • ἔμμορε
      has a share[fn]mid. εἵμαρ-το was apportioned[/fn]
       
       
    • τέτροφε
      (τρᾰφ-)
      is thickened
       
    • ἐπι-δέδρομε
      runs over
       
       
    • δέδορκε
      sees
       
    • ἔοργας
      has done
       
    • ἔολπα
      I hope

    Weak forms: πεπαρ-μένος pierced, τέτραπ-το (τρέπ-ω), ἐ-τέταλ-το (τέλλω).

    But ερ-, ελ- in ἐερ-μένος strung (Lat. sero), ἔρχ-αται are packed in, part. ἐεργ-μέναι (ϝέργ-ω), and ἐελ-μένος cooped in: cp. § 31.6.

    ρῑ appears in βέβρῑθε is heavy, ἔρρῑγα dread, πεφρῑκ-υῖαι bristling, τετρῑγ-υῖαι chirping, with no corresponding weak stem. In these words ρῑ seems to come from original ερ, ρ, or ; cp. § 29.4.

  7. ον and ᾰ (for )
     

    • γέγονε
      1st plur. γέγᾰ-μεν
      is born
       
    • πέπονθα
      2nd plur. πέπασθε (for πεπαθ-τε)
      part. πεπᾰθ-υῖα
      I suffer
       
    • μέμονας
      2nd plur. μέμᾰ-τε
      art eager
       
    • λελόγχ-ᾰσι
      aor. ἔλᾰχ- ον
      have as portion
       
    • πέφᾰ-ται
      φόν-ος
      is slain
       
    • τέτᾰ-ται
      τόν-ος
      is stretched
       
    • δεδα-ώς (§ 31.5)

    But we find αν in κεχανδ-ώς containing (aor. ἔχᾰδε).

  8. ο and ε
    • τέτοκα (Hes. Op. 591[fn]Cp. aor. ἔ-τεκ-ον.[/fn]
       
       
    • δέδεγ-μαι[fn]Cp. προ-δοκ-αί ambush.[/fn]
      I await
       
    • ἕσ-σαι
      are clothed
       
    • ἀνήνοθεν
      mounted up[fn]Of a stream of blood, Il. 11.266.[/fn]
       
    • ἐπ-ενήνοθε
      is upon
       
    • ἀγηγέρ-ατο[fn]Cp. ἀγορ-ή[/fn]
      were assembled
       
    • κεκοπ-ώς
      striking

    Properly the form with ο should interchange with a form without a vowel (τοκ- with τκ-, etc.), but when this is impossible ε remains in the weak stem: see § 6.6.

    ἀνήνοθε answers in meaning to the Attic ἀνθέω, to be on the surface, come forth upon: the present would be ἀνέθ-ω (related to ἄνθ-ος as ἀλέγ-ω to ἄλγ-ος). So ἐν-ήνοθε supposes ἐνέθ-ω, weak form ἐνθ-.

  9. Stems which take the suffix κ*.[fn]A word may be said here on the origin of the perfects in -κᾰ. They may be regarded as formed in the ordinary way from stems in which a root has been lengthened by a suffixed κ, as in ὀλέ-κ-ω, ἐρύ-κ-ω (§ 45), πτήσσω (for πτη-κ-ι̯ω, cp. ἔ-πτα-κ-ον), δειδίσσομαι (for δει-δϝικ-ι̯ο-μαι). Thus ὀλώλεκα is the regular perfect of ὀλέκω, and πέπτωκα, δείδοικα, answer to the weak stems πτᾰ-κ-, δϝῐ-κ-. So βέβηκα, ἕστηκα answer to (possible) presents *βή-κω (cp. βάκ-τρον), *στή-κω. It is not necessary to suppose an actual stem in κ in each case; a few instances would serve to create the type. The reason for the use of the longer stems βη-κ, στη-κ, etc., was probably that the forms given by the original stems were too unlike other perfects. The characteristic -ᾰ would be lost by contraction with the preceding vowels. It is a confirmation of this view that the stem with -κα is in the same form as the present stems with a suffixed κ, γ, θ (§ 45), or σκ (§ 48). A similar theory may be formed of the perfects in -θα, of which the germs have been mentioned above. βέβρω-θα is related to a part. βεβρω-ώς (§ 26.4) as τέθνη-κα to τεθνη-ώς, and to a mid. *βέβρω-μαι (cp βεβρώ-σεται) as βέβλη-κα to βέβλη-μαι. If in a few more cases, such as βέβρῑ-θα (βρῐ-), εἴωθα (suē-tus), γέγη-θα (γαϝ-ιω), we had had short forms of the stem without -θ-, the suffix -θα would have been felt to characterize the perfect active; that is to say, the type of the "Perfect in -θα" would have been created, and might have spread as the perfect in -κα has done.
    The aorists in -κᾰ are to be accounted for in the same way. The κ may be traced in the present δώκω (on the inscription of Idalion, see Curt. Stud.vii. 243) and in the noun θήκ-η, which points to a verb stem θη-κ-. [/fn]

    When the stem ends in a vowel, certain forms of the perfect active take κ, thus filling the hiatus which would otherwise be made between the stem and the ending: as in ἕστη-κ-ας, δείδοι-κ-α, τεθαρσή-κ-ᾱσι. The perfects of this type–including those of which no forms with κ are actually found–may be divided again into

    (a) perfects with variable root vowel

    • ἕστηκα
      1st plur. ἕστᾰ-μεν
      I stand
    • δείδοικα
      1st plur. δείδῐ-μεν
      I fear
       
    • πέφυκε
      3rd plur. πεφύ- ᾱσι
       
    • βέβηκα
      infinitive βεβᾰ́μεν
       
    • τέθνηκα
      imperative τέθνᾰ-θι
    • τέτληκα
      imperative τέτλᾰ-θι

    Add also μέμῡ-κε is closed (of a wound), δέδῡ-κε is sunk in, though the short form is not found.

    (b) Perfects with invariable long vowel, especially η and ω (discussed in § 14)

    • βεβλή-κ-ει
      mid. βέβλη-ται[fn]Cp. ξυμ-βλή-την, βλή-μενος[/fn]
      struck
       
    • κέκμη-κ-ας
      are weary
       
       
    • πεπλη-μένος
      brought near
       
    • κέκλη-μαι
       
       
    • εἴρη-ται
       
    • μέμνη-μαι
       
    • τετμη-μένος
       
    • βεβρω-κ-ώς
      fut. mid. βεβρώ-σεται
      having eaten
       
    • μέμβλω-κ-ε
      is gone
    • πεπρω-μένος
      fated

    Similarly, from disyllabic stems, δεδάη-κε (aor. ἐ-δάη-ν) has learned (Od. 8. 134), τετύχη-κε (Od. 10. 88), and the participles κεχαρη-ότα (ἐ-χάρη-ν), βεβαρη-ότα, κεκαφη-ότα, τετιη-ότες.

    To this class belong the perfects of derivative verbs in -αω, -εω, -οω, -υω, as βεβίη-κ-εν (Il. 10.145, 172; 16.22), ὑπ-εμνήμυ-κε (Il. 22.491), δεδειπνή-κει (Od. 17.359), τεθαρσή-κ-ᾱσι (Il. 9.420, 687): κεκοτη-ότα, κεκορη-ότα, ἀκ-άχη-μαι, ἀλ-άλη-μαι, ἀλα-λύκτη-μαι.

    παρ-ῴχη-κεν (Il. 10.252, with v.l. παρῴχωκεν) is formed as if from *παρ-οιχέω, for παρ-οίχομαι.

    ἀδη-κ-ότες (Od. 12.281, and four times in Il. 10) means displeased, disgusted, and should probably be written ἀαδηκότες, from ἀαδέω (for ἀ-σϝᾰδ-εω).

    The subjunctive ἱλήκῃσι (Od. 21.36), optative ἱλήκοι (H. Apoll. 165) point to a perfect ἵληκα or present ἱλή-κω.

  10. A perfect in -θα may be recognized in ἐγρηγόρ-θᾱσι keep awake(Il. 10. 419): perhaps in the optative βεβρώθοις (Il. 4. 35).

    In general the perfects of derivative verbs are formed with an invariable stem.

    κεκορυθ-μένος, πεπόλισ-το, ὀδώδυσ-ται, κεκονῑ-μένος

    But no such perfects are used in the active.

23.The reduplication takes the following forms:

  1. An initial consonant is repeated with ε. This is the general rule: we need only notice the perfects in which an original consonant has been lost.

    A labial semi-vowel (ϝ) in ἐ-ελ-μένος cooped in (for ϝε-ϝελ-μένος), εἰρύαται (ϝερυ-) are drawn up, εἴλυ-το (ϝελυ-, volvo), ἔ-οργα (ϝέργ-ον), ἔ-ολπα, ἔοικα, mid. ἤϊκ-το (unless this comes from ἐΐσκω).
    A sibilant (σ) in ἕ-στηκα (for *σέ-στηκα), ἐ-ερ-μένος strung together (Latin sero). But the σ is retained in σέσηπε.

  2. Stems beginning with two consonants (except when the second is ρ, λ, μ, or ν), or with ζ, usually prefix ε- only.

    δι-έ-φθορας, ἐ-φθίατο, ἐ-κτῆσθαι (but κέκτημαι, Hes. Op. 437), ἐ-ζευγμέναι

    But we find πε-πτηώς, πέ-πτανται. And in ἕστηκα the rough breathing represents original σ.

    The group σϝ has been lost in ἑ-ᾱδώς (either σε-σϝᾱδώς or ἐ-σϝᾱδώς) pleasing, and εἴωθα, ἔωθα (Lat. suē-sco).

    The group δϝ has the effect of lengthening the vowel of the reduplication in δείδοικα, δείδι-μεν, etc., which represent original δέ-δϝοι-κα, δέ-δϝῐ-μεν, etc.

    Initial ρ-, which generally stands for ϝρ- (sometimes σρ-), gives ἐρρ-, as in ἔρρηκται (ϝρηγ-), ἐρρίζωται. Sometimes εἰρ-, as εἴρηται (ϝρη-, cp. ver-bum), and εἰρύαται (ῥύομαι, ϝρῡ- protect). One stem reduplicates ρ-, viz. ῥε-ρυπωμένα, from ῥυπόω.

    Similarly we have ἔμμορε, mid. εἴμαρ-ται (σμαρ-), and ἔσσῠται (σεύω, root κι̯ευ-: also εἴληφα (post. Hom., cp. ἔλλαβον, § 67.)

    We must distinguish between (1) phonetic loss, as of σ or ϝ, and (2) substitution of initial ἐ- for the reduplication. The latter may be seen (e.g.) in ἐ-κτῆσθαι, which cannot be derived by phonetic decay from κε-κτῆσθαι. The distinction will serve to explain the difference between εἵμαρται, which is the proper representative of an original σέ-σμαρ-ται, and ἔμμορε, which follows the general tendency to double an initial μ-, ν-, λ- or ρ- after the augment.

  3. Attic Reduplication; as ὄπ-ωπα have seen, ἐλ-ήλα-το was driven, ἐγρ-ήγορα I am awake.

    The syllable which follows the Attic reduplication may vary in quantity.

    ἄρηρε, fem. part. ἀρᾰρυῖα
    ἐρήριπε, mid. ἐρέριπτο

    Usually it is long.

    ἐλήλαται, ἀρηρομένος, ἀκηχεμένος, ὀδώδυσται, ἠρήρειστο, ἐρήρισται (Hes. fr. 219)
    3rd plur. ἀγηγέρατο, ἐρηρέδαται, ὀρωρέχαται

    But it is short in ἀκάχημαι, ἀλάλημαι.

  4. Temporal Augment (see § 67).

    ἐφ-ῆπ-ται (ἅπτω)
    κατ-ῄκισ-ται (αἰκίζω)
    ἤσκη-ται (ἀσκέω)
    ᾐσχυμμένος

  5. In a few cases there is no reduplication.

    οἶδα, for ϝοῖδα, Sanskrit veda
    ἔρχ-αται are shut in (ϝεργ-), pluperfect ἔρχ-ατο and (with augment) ἐέρχατο
    εἷμαι am clothed with (ϝεσ-), ἕσ-σαι, pluperfect ἕσ-σο, ἕσ-το and (with augment) ἕ-εσ-το, dual ἕσ-θην, 3rd plur. εἵατο, part. εἱμένος

    Reduplication is not to be found in the ει- of εἷμαι, εἱμένος, since these are for ϝέσ-μαι, ϝεσ-μένος (as εἷμα for ϝέσμα). The 3rd singular perfect occurs once in Homer, in Od. 11.191, where the best MSS. have ἧσται, others εἷσται and εἷται. The true form is probably ἕσται, preserved in an oracle in Hdt. 1.47 (cp. ἕσσαι).

    ἀμφιαχυῖα (Il. 2.316) crying around can hardly be divided ἀμφ-ιαχυῖα, since the stem ἰᾰχ- has initial ϝ- (§ 390). But a stem ϝηχ- (ϝηχή cry), weak form ϝᾰχ-, without reduplication would give the feminine participle ϝᾰχυῖα, whence ἀμφι-αχυῖα.

    These examples make it doubtful whether initial ϝ- was originally reduplicated in the perfect stem. In Sanskrit the roots which begin with va- (answering to Greek ϝε-) take u-, as uvâca (vac-, Greek ϝεπ-). Thus the ϝε- of ϝέϝοικα, ϝεϝελμένος, etc. may be later, due to the analogy of other perfects.

    δέχ-αται await (Il. 12.147), pluperfect ἐ-δέγμην (Od. 9.513, 12.230), part. δέγμενος (Il. 2.794, 9.191, 18.524; Od. 20.385), with the same perfect meaning that we have in δέδεγμαι (await, not receive, § 28): while in other places ἔ-δεκτο, etc., are no less clearly aorists. It seems that we must recognize a perfect form *δέγμαι (Buttm. G. G. ii. 149, Curt. Verb. ii. 144), probably older than δέδεγμαι.[fn]With the instances here given we may place the Cretan καταϝελμένοι, which occurs in the inscription of Gortyn with the meaning gathered together, assembled (cp. Homeric ἐελμένοι crowded). Baunuck, however, takes it for καταϝηλμένοι, supposing loss of ϝ and contraction from καταϝεϝελμένοι.[/fn]

  6. The reduplication in δει-δέχ-ᾰται they welcome, seems to be that of the "Intensive" forms, as in δει-δίσκομαι: see § 61. The form belongs to δείκ-νυμι, not δέχ-ομαι (see Veitch).

24. In the 3rd plural

  1. The long stem with -ᾱσι (-α-ΝΤΙ) is comparatively rare.

    πεποίθᾱσι (Il. 4.325)
    ἑστήκᾱσι (Il. 4.434, v.l. ἑστήκωσι)
    κατα-τεθνήκᾱσι (Il. 15.664)
    τεθαρσήκασι (Il. 9.420, 682)
    ἐγρηγόρθᾱσι (Il. 10. 419).

    These forms evidently result from generalising the stem in α. So we have οἶδα-ς (Od. 1.337), οἴδᾰ-μεν, οἴδασι in Herodotus (and in Attic, see Veitch s. v.).

  2. The final consonant of the stem, if a labial or guttural, is aspirated before the -ᾰται, -ᾰτο of the middle.
     

    • ἐπι-τετράφ-αται
      are entrusted
       
    • τετράφ-ατο
      were turned
       
    • ἔρχ-αται (ϝεργ-)
      are shut in
       
    • ὀρωρέχ-αται (ὀρέγ-ω)
      are stretched out
       
    • δειδέχ-αται (δείκ-νυμι)
      welcome
    • κεκρύφ-αται (Hes. Op. 386).

    The aspirated forms of the active, such as εἴληφα, κέκοφα, are entirely unknown to Homer.

    It has been pointed out by Joh. Schmidt (K. Z. xxviii. 309) that the aspiration in these cases is due to the analogy of the forms in which a similar aspiration is caused by the ending: τετράφ-αται because of the 2nd plural τέτραφ-θε, infinitive τετράφ-θαι. This explains why a final dental is not affected: for δ before θ passes into σ.

  3. An anomalous ε for ῐ appears in δει-δέχ-αται (δείκ-νυμι, see § 23.6), ἐρ-ηρέδ-αται (ἐρείδω, cp. ἠρισ-μένος Hesych.), and ἀκ-ηχέδ-αται (ἀκαχίζω).
  4. A final δ of the stem sometimes appears only in the 3rd plur.: as ἀκηχέδ-αται, ἐρράδ-αται (ῥαίνω, 1st aor. ῥάσσατε), ἐληλάδ-ατο. But the last of these forms is doubtful; it occurs only in Od. 7.86 χάλκεοι μὲν γὰρ τοῖχοι ἐληλάδατʼ, where some good MSS. have ἐρηρέδατ᾿.

25. Interchange of Stems. The original variation between the strong and the weak form is disturbed by various causes.

  1. The ο-form of the stem is found instead of the weak form in

    εἰλήλουθ-μεν (for εἰλήλυθ-μεν)
    we are come

    ἄωρτο (cp. ἄερ-θεν)
    was hung aloft

    ἐγρήγορθε (with the inf. ἐγρήγορθαι, Il. 10.67; cp. ἐγρηγορτί 10.182)
    keep awake

    ἄνωγμεν (H. Apoll. 528)

    cp. ἔοιγμεν (in tragedy), δέδοιγμεν (Et. M.).

  2. The strong stem of the present takes the place of the weak stem in

    συν-έρρηκται (Attic ἔρρωγα)
    λέλειπ-ται
    ἐζευγ-μέναι
    ἠρήρειστο (ἐρείδω)

    also in ἐερ-μένος, ἐελ-μένος, ἔρχ-αται (§ 22.6). So κεχανδ-ώς (for κεχᾰδ-ϝώς, χανδάνω).

    ἕστητε, commonly read in Il. 4.243, 246, is an error for ἔστητε : see § 76.

  3. The influence of the present may further be traced in the perfects which take ῑ for ει (§ 22.4), and ῡ, ευ for ου (§ 22.5). So ἐδηδ-ώς (but ἐδωδή), προ-βέβουλα (βούλομαι).

    In all these cases it is worth noticing that the change does not affect the metrical form of the word: e.g. we may read εἰλήλυθμεν, ἔρρακται, ἐζυγμέναι, ἠρήριστο, etc., and some of these may be the true Homeric forms.

    The weak stem appears to take the place of the ο-form in δείδια (as to which see § 22.4), and in ἀνα-βέβρῠχεν (Il. 17.54) gushes up. For the latter Zenodotus read ἀναβέβροχεν–doubtless rightly, since this is the correct perfect of ἀνα-βρέχω.

    In Attic reduplication the second vowel of a disyllabic stem may be short, as in ἐλήλυθα (less common in Homer than εἰλήλουθα), and κατερήριπε (Il. 14.55).

Suggested Citation

D.B. Monro, A Grammar of the Homeric Dialect. Carlisle, Pennsylvania: Dickinson College Commentaries, 2014. ISBN: 978-1-947822-04-7. https://dcc.dickinson.edu/grammar/monro/perfect