{"slug": "face-to-face", "name": "FACE TO FACE", "meter_label": "8.7.8.7 with Refrain", "meter": {"pattern": [8, 7, 8, 7], "doubled": false, "refrain": true, "alleluia": false, "irregular": false}, "composer": "grant-colfax-tullar", "composer_date": "1899", "canonical_key": {"tonic": "bes", "mode": "major"}, "preferred_harmonization": "face-to-face-tabernacle-1921", "provenance": {"layer": "tune", "source": "tabernacle-hymns-no-2-1921", "source_kind": "hand_encoded", "pd_basis": "pub_pre_1931", "pd_evidence": "Music by Grant Colfax Tullar, read for this record off the dated printing in Tabernacle Hymns, No. 2 (Chicago: Tabernacle Publishing Co., 1921), hymn 62, \"Face to Face.\" \u2014 archive.org tabernaclehymnsn00chic leaf 0070, collated against a second scan of the same plate, archive.org tabernaclehymnsn02chic leaf 0068. That page credits \"Grant Colfax Tullar.\" at the right of the title and \"Mrs Frank A. Breck.\" at the left, marks the piece \"Moderato.\", and foots it \"Copyright, 1899 by Tullar-Meredith Co   Used by per\" \u2014 the footer line runs off the trim on both scans and is quoted only as far as it prints. US public domain: the underlying copyright is 1899 and the printing actually read is 1921, both well before 1931. NEITHER PRINTING PRINTS A TUNE NAME; FACE TO FACE is the name under which the tune is universally catalogued and is taken from the hymn's own title as the page heads it. THE METRE WAS TAKEN FROM THE MUSIC, not from any heading: the verse sets four lines of 8.7.8.7 over thirty soprano noteheads (two systems of fifteen, strictly one note to a syllable) and the chorus adds thirty more, so 8.7.8.7 with Refrain. Not a duplicate: tools/tune_fingerprint.py, run on a bare soprano MIDI written from THIS page reading and from nothing else, scores 0.58 against its nearest corpus neighbour (EWING) \u2014 far below the 0.80 duplicate threshold. No Hymnary cross-check was available: Hymnary's /tune/ route returns 403 to this repo.", "pd_jurisdictions": {"US": true}}, "harmonization": {"slug": "face-to-face-tabernacle-1921", "tune": "face-to-face", "music_file": "face-to-face-tabernacle-1921.ly", "canonical_key": {"tonic": "bes", "mode": "major"}, "time": "4/4", "pickup": null, "system_breaks": [15, 30, 45], "breaks_manual": true, "final_barline": "|.", "amen": false, "verse_variable_melisma": false, "refrain": {"soprano_note_start": 31}, "markings": [{"soprano_note": 31, "kind": "refrain"}], "harmonizer": "grant-colfax-tullar", "arrangement_type": "harmonization", "source_line": "Setting: Grant Colfax Tullar, 1899, as printed in Tabernacle Hymns, No. 2, 1921. Public domain.", "provenance": {"layer": "harmonization", "source": "tabernacle-hymns-no-2-1921", "source_kind": "hand_encoded", "pd_basis": "pub_pre_1931", "pd_evidence": "Four-part setting of Tullar's FACE TO FACE as printed in Tabernacle Hymns, No. 2 (Chicago: Tabernacle Publishing Co., 1921), hymn 62, \"Face to Face.\" \u2014 the leaf carries no printed page number \u2014 headed \"Mrs Frank A. Breck.\" (left) and \"Grant Colfax Tullar.\" (right), marked \"Moderato.\", in B-flat major and 4/4 with no pickup, the word \"Chorus.\" standing at the head of the third system, and footed \"Copyright, 1899 by Tullar-Meredith Co   Used by per\". THAT FOOTER LINE RUNS OFF THE TRIM ON BOTH SCANS and is quoted above exactly as far as it prints; it is not a full transcription of the copyright notice, which the binding cuts. US public domain: the underlying copyright is 1899 and the printing read is 1921, both well before 1931, and no modern arranger stands anywhere in the chain. TWO DATED SCANS OF THAT PLATE WERE READ AND REQUIRED TO AGREE \u2014 archive.org tabernaclehymnsn00chic leaf 0070 (2365 px across, staff space 28.54-28.83 px, median 28.67, i.e. 14.3 px per diatonic step) and archive.org tabernaclehymnsn02chic leaf 0068 (1877 px across, staff space 22.57-22.89 px, median 22.79, i.e. 11.4 px per step); the two items are two scans of the one book, and at this hymn n00chic runs two leaves LATER than n02chic, the same offset the ADELAIDE record notes. THE NOTEHEADS WERE MEASURED FIRST AND THEN READ BY EYE, and it is the eye reading that ships. Per scan, with its own binarisation and its own five-line fits: ink is whatever is darker than 82% of the LOCAL paper level (a 3x3 grey opening, then a 41x41 grey dilation, then a 41x41 box filter), closed 5x5; holes under 0.60 staff space squared, no wider than 1.30 and no taller than 0.80 staff space are filled, which turns each hollow half or dotted-half OUTLINE into a disc without flooding the boxes that staff lines and stems enclose; staff lines are suppressed by vertical run length AFTER that fill; and every head is converted to a diatonic step against a five-line fit INTERPOLATED FROM BAND-BY-BAND MEASUREMENTS around that head's own x, never from a straight fit across the system. Whole-system grid residuals: 0.34-1.28 px mean against a 22.7-28.9 px staff space. Column-table residuals: 0.203 step mean on scan 1 system 1, 0.153 system 2, 0.158 system 3, 0.188 system 4; 0.131, 0.134, 0.137 and 0.191 for the same four systems on scan 2. EVERY ONE OF THE 240 NOTE POSITIONS WAS THEN CONFIRMED BY EYE off grid-overlay crops at 4x to 12x. The eye overruled the machine three times and the second scan, read by eye, agreed with the correction each time: the alto of system 3 note 13 is E4 (machine +1 = F4), the bass chord of system 2 note 11 is B-flat3 over F3 (scan-2 machine +5/+3 = E-flat3 over C3), and the tenor of system 3 note 9 is A3 (scan-2 machine +7 = G3). The whole of SYSTEM 1 was reproduced independently by the scan-2 machine pass \u2014 15 columns, 60 note positions, mean residual 0.131 step \u2014 and matches this encoding in all four voices with 0 disagreements; systems 2, 3 and 4 were read by eye on scan 1 and collated column by column against the scan-2 machine pass, with a scan-2 eye crop taken at every disagreement. ACCIDENTALS WERE SWEPT SEPARATELY AND BY EYE over every column of BOTH leaves; the automated figures above see staff positions only and never covered accidentals. The two-flat signature stands at the head of every staff. The sweep found exactly FOUR printed accidentals and no others: a SHARP before the alto C4 and a NATURAL before the bass E3 at soprano note 24 (system 2, \"Je-\"); the same pair again at soprano note 54 (system 4, \"shall\"); and a NATURAL before the alto E4 at soprano note 42 (system 3, \"star-\"). THE MEASURE RULE WAS THE TRAP AND IT IS APPLIED DELIBERATELY: in systems 2 and 4 the very next chord prints the same G4-over-C4 pair with no second sharp and the same B-flat3-over-E3 pair with no second natural, inside the same bar, so those are C sharp 4 and E natural 3 too. Two independent witnesses on the one plate settle it \u2014 systems 2 and 4 are musically identical here and BOTH print exactly one sharp and one natural followed by one unmarked repetition \u2014 and the harmony confirms it: the chord is E-G-B flat-C sharp, a fully diminished seventh on E (vii dim 7 of the dominant), resolving E to F, C sharp to D and G to F onto the B-flat-over-F that opens the next bar. Had the second chord been C natural over E flat it would have been a C minor seventh, which does not resolve that way. The system-3 natural likewise carries from soprano note 42 to notes 43 and 44 (the alto E4 never moves), giving C7 three times over a C3 pedal into F major; it EXPIRES at the next barline, so the alto E in that last bar is E flat again, which is what makes the chord F7 leading into system 4. EVERY CHORD WAS NAMED. All sixty are diatonic B-flat major or one of three secondary dominants (C7 = V7/V, the E diminished seventh above, and the F7 that is simply V7). Two that surprised the reading and are NOT errors: under \"Sav\" (system 1 bar 2) the plate reads bass C3, tenor F3, alto E-flat4, soprano B-flat4 \u2014 C minor seventh (ii7) with the tenor F standing as a fourth over the bass C and rising to G, which is how the chord completes itself on the second quarter \u2014 and it resolves on \"-iour,\" where the soprano falls to A4 against alto E-flat4, an A4/E-flat4 tritone that is simply the third and seventh of F7. The tenor moves F3-G3 under that suspension, slurred on the plate, making C minor seventh on the second beat. NO SCAFFOLD WAS USED AND NOTHING WAS CHECKED AGAINST ONE: no MIDI, no .nwc, no Cyber Hymnal asset and no existing encoding was opened at any point. No Hymnary incipit cross-check was available either \u2014 Hymnary's /tune/ route returns 403 to this repo \u2014 and none was used. The one file written from this reading was a bare soprano MIDI used for the duplicate guard, which scores 0.58 against its nearest corpus neighbour (EWING), far below the 0.80 threshold. DEPARTURES FROM THE PAGE, in full: (1) TWO FERMATAS ARE NOT ENGRAVED. The plate prints a fermata above the treble staff and an inverted fermata below the bass staff over the last chord of system 3 (soprano note 45, the held \"sky;\") and again over soprano note 53 in system 4 (the held \"-ry,\"). This corpus cannot engrave a fermata, so both are dropped and recorded here. (2) The \"Moderato.\" tempo word is not carried, this corpus holding no tempo marks. (3) AT THE LAST CHORD OF SYSTEMS 2 AND 4 THE PLATE PRINTS FIVE NOTES, not four: the lower staff carries three hollow dotted heads on one down stem, B-flat3 over F3 over B-flat2, under treble B-flat4 and D4. The four voices taken here are the four the voice-leading of the preceding F7 gives \u2014 soprano B-flat4 (from C5), alto D4 (from E-flat4), tenor B-flat3 (the leading tone A3 rising), bass B-flat2 (from F3) \u2014 and the added F3 that fills in the fifth is dropped, because this corpus admits exactly four voices and forbids divisi. That the four voices are right is confirmed by the end of system 1, where the soprano has F4 instead of B-flat4 and the same voice-leading yields a complete triad, and the plate there prints exactly four notes with no added F3. (4) The printed ellipsis after \"sky;\" in the chorus's first lyric line, which marks the held note, is not carried. Nothing else departs: every pitch, every duration, every rest, all four accidentals, the tenor slur in bar 2 and the closing light-heavy double bar are the plate's, and the systems break exactly where the page breaks them.", "pd_jurisdictions": {"US": true}, "encoding_origin": "Hand-encoded from the page images of Tabernacle Hymns, No. 2 (1921), hymn 62 \u2014 archive.org tabernaclehymnsn00chic leaf 0070, collated notehead by notehead against archive.org tabernaclehymnsn02chic leaf 0068. No MIDI, no Cyber Hymnal asset and no existing encoding was used at any point.", "encoding_license": "PD-dedication"}, "music_ly": "%% FACE TO FACE (8.7.8.7. with Refrain) \u2014 Grant Colfax Tullar, 1899. Four SATB\n%% voice streams, canonical key B-flat major, 4/4, NO pickup, sixteen full bars:\n%% four systems of four bars, fifteen soprano noteheads to a system, sixty in all.\n%% The chorus is systems 3-4, i.e. soprano notes 31-60.\n%% RULES (CI-enforced): exactly the variables soprano/alto/tenor/bass/global;\n%% absolute octaves; each voice rhythmically complete; NO chords <>, NO <<>>,\n%% NO \\secondShift/\\unisonWholeShift/tightSpacing, NO \\break, NO \\bar, NO lyrics.\n%%\n%% SOURCE OF THE NOTES. Tabernacle Hymns, No. 2 (Chicago: Tabernacle Publishing\n%% Co., 1921), hymn 62, \"Face to Face.\" \u2014 the leaf carries no printed page\n%% number \u2014 headed \"Mrs Frank A. Breck.\" on the left and \"Grant Colfax Tullar.\"\n%% on the right, marked \"Moderato.\", in B-flat major (two flats) and 4/4 with no\n%% pickup, and footed \"Copyright, 1899 by Tullar-Meredith Co   Used by per\" \u2014\n%% that footer line RUNS OFF THE TRIM on both scans and is quoted here exactly as\n%% far as it prints. Read off TWO independent scans of that same plate:\n%% archive.org tabernaclehymnsn00chic leaf 0070 (2365 px across, staff space\n%% 28.54-28.83 px, median 28.67 = 14.3 px per diatonic step) and archive.org\n%% tabernaclehymnsn02chic leaf 0068 (1877 px across, staff space 22.57-22.89 px,\n%% median 22.79 = 11.4 px per step). The two items are two scans of the one book,\n%% and at this hymn n00chic runs two leaves LATER than n02chic \u2014 exactly as it\n%% does at ADELAIDE, whose record notes the same offset.\n%%\n%% HOW THE NOTEHEADS WERE READ. Per scan, with its own binarisation and its own\n%% five-line fits: ink is whatever is darker than 82% of the LOCAL paper level (a\n%% 3x3 grey OPENING, then a 41x41 grey dilation, then a 41x41 box filter),\n%% followed by a 5x5 binary closing; holes smaller than 0.60 staff space squared\n%% and no wider than 1.30 / taller than 0.80 staff space are filled, which turns\n%% each hollow half or dotted-half OUTLINE into a disc without flooding the boxes\n%% that staff lines and stems enclose; staff lines are suppressed by vertical run\n%% length AFTER that fill; every head is then converted to a diatonic step\n%% against a five-line fit INTERPOLATED FROM BAND-BY-BAND MEASUREMENTS in a\n%% window around that head's own x, never from a straight fit across the system.\n%% Whole-system grid residuals came out 0.34-1.28 px mean against a 22.7-28.9 px\n%% staff space. THE MACHINE PASS IS A FINDER, NOT A DECIDER: every pitch below\n%% was then confirmed BY EYE off grid-overlay crops at 4x to 12x, and it is the\n%% eye reading that ships. It overruled the machine three times \u2014 the alto of\n%% system 3 note 13 (E4, machine +1), the bass chord of system 2 note 11 (B-flat3\n%% over F3, machine +5/+3 on scan 2), and the tenor of system 3 note 9 (A3,\n%% machine +7 on scan 2) \u2014 and in each case the second scan, read by eye, agreed\n%% with the correction.\n%%\n%% THE TWO SCANS AGREE. The whole of system 1 was reproduced independently on\n%% scan 2 by the machine pass (15 columns, 60 note positions, mean residual 0.131\n%% step) and it matches this encoding in all four voices with 0 disagreements.\n%% Systems 2, 3 and 4 were read by eye on scan 1 and collated column by column\n%% against the scan-2 machine pass plus scan-2 eye crops at every disagreement.\n%%\n%% ACCIDENTALS ARE NOT WHAT A NOTEHEAD PROBE SEES. Every column of BOTH leaves\n%% was swept separately BY EYE at high magnification; the figures above cover\n%% staff positions only and never covered accidentals. The two-flat signature\n%% stands at the head of every staff. The sweep found exactly FOUR printed\n%% accidentals in the body of the hymn, and no others anywhere:\n%%   - system 2, soprano note 24 (\"Je-\"): a SHARP in the treble before the alto\n%%     head on the first ledger line below the staff (C sharp 4), and a NATURAL\n%%     in the bass staff before the bass head in the second space (E natural 3);\n%%   - system 4, soprano note 54 (\"shall\"): the same pair again, at the same\n%%     place in the parallel bar.\n%%   - system 3, soprano note 42 (\"star-\"): a NATURAL in the treble before the\n%%     alto head on the bottom line (E natural 4).\n%% THE MEASURE RULE WAS THE TRAP HERE and it is applied deliberately. In systems\n%% 2 and 4 the VERY NEXT chord prints the same G4-over-C4 pair with NO second\n%% sharp and the same B-flat3-over-E3 pair with no second natural; both stand\n%% inside the same bar, so they are C sharp 4 and E natural 3 as well. Two\n%% independent witnesses on the one plate settle it: systems 2 and 4 are\n%% musically identical here and BOTH print exactly one sharp and one natural\n%% followed by one unmarked repetition. The chord is E-G-B-flat-C sharp, a fully\n%% diminished seventh on E \u2014 vii(o)7 of the dominant \u2014 resolving with E to F,\n%% C sharp to D and G to F onto the B-flat-over-F that opens the next bar; had\n%% the second chord been C natural / E flat it would have been a C minor seventh,\n%% which does not resolve that way and which the plate would have had to mark.\n%% The same rule carries the system-3 natural forward from soprano note 42 to\n%% notes 43 and 44 (the alto E4 does not move), giving C7 - C7 - C7 over a C3\n%% pedal into the F major of the next bar; it EXPIRES at that barline, so the\n%% alto E in bar 12 is E flat again, which is what makes that chord the F7 that\n%% carries into system 4.\n%%\n%% NO SCAFFOLD WAS USED AND NOTHING HERE WAS CHECKED AGAINST ONE. No MIDI, no\n%% .nwc, no Cyber Hymnal asset, no Hymnary incipit and no existing encoding was\n%% opened at any point. (Hymnary's /tune/ route is behind a 403 for this repo, so\n%% no incipit cross-check was available; none was used.) The only file written\n%% from this reading was a bare soprano MIDI used for the duplicate guard, which\n%% scores 0.58 against its nearest corpus neighbour (EWING), far below the 0.80\n%% threshold.\n%%\n%% RHYTHM, as the plate prints it. Every system is four bars of 4/4. Bars 1, 5,\n%% 9 and 13 are a dotted quarter and five eighths. Bars 2 and 10 are a half plus\n%% (bar 2) a dotted quarter and an eighth rest, (bar 10) a second half. Bars 6\n%% and 14 are a half, a quarter and two eighths. Bars 3 and 11 are a dotted\n%% quarter and five eighths; bars 7 and 15 are two quarters, a dotted quarter and\n%% an eighth. Bars 4, 8 and 16 are a dotted half and a quarter rest in all four\n%% voices. Bar 12 is the one bar where the outer voices hold and the inner voices\n%% move: soprano and bass sustain a dotted half (each under a printed fermata)\n%% while alto and tenor walk three quarters, all four then resting a quarter.\n%%\n%% TWO PLACES WHERE THE PLATE DOES SOMETHING THE VOICE COUNT MUST ABSORB:\n%%   (1) BAR 2, the tenor alone moves under the held syllable \"Sav-\": it sings\n%%       F3 then G3 as two quarters against the soprano's and alto's half note,\n%%       and the plate draws a SLUR over the pair. That slur is reproduced. The\n%%       chord on the second quarter is C3-G3-E flat 4-B flat 4 = C minor seventh.\n%%   (2) BAR 2 note 8 and the whole of bar 3's bass, the plate prints ONE head\n%%       carrying both an up and a down stem where tenor and bass are in unison.\n%%       Those are written here as the same pitch in both voices, which is what\n%%       the double stem means.\n%%\n%% DEPARTURES FROM THE PAGE, in full \u2014 see the harmonization yaml, which states\n%% them again: (1) the two printed fermatas (over and under the last chord of\n%% system 3, over and under soprano note 53 in system 4) are NOT engraved, this\n%% corpus having no fermata; (2) the \"Moderato.\" tempo word is not carried, this\n%% corpus holding no tempo marks; (3) at the last chord of systems 2 and 4 the\n%% plate's lower staff prints THREE hollow heads, B-flat3 over F3 over B-flat2,\n%% a five-note chord \u2014 the four voices taken here are the four the voice-leading\n%% gives (soprano B-flat4, alto D4, tenor B-flat3 from the leading tone A3, bass\n%% B-flat2 from F3), and the added F3 that fills the fifth is dropped, because\n%% this corpus admits four voices and forbids divisi; (4) the printed ellipsis\n%% after \"sky;\" in the chorus's first lyric line, which marks the held note, is\n%% not carried. Nothing else departs: every pitch, every duration, every rest,\n%% all four accidentals, the tenor slur in bar 2 and the closing light-heavy\n%% double bar are the plate's, and the plate's own four barlines a system are the\n%% barlines here.\n\\version \"2.24.0\"\n\nglobal  = { \\key bes \\major \\time 4/4 }\n\nsoprano = {\n  f'4. f'8 g'8 f'8 d'8 bes'8 |\n  bes'2 a'4. r8 |\n  c''4. bes'8 a'8 g'8 f'8 g'8 |\n  f'2. r4 |\n  f'4. f'8 g'8 f'8 d'8 bes'8 |\n  c''2 g'4 g'8 g'8 |\n  f'4 d''4 d''4. c''8 |\n  bes'2. r4 |\n  c''4. c''8 d''8 c''8 bes'8 a'8 |\n  bes'2 d''2 |\n  c''4. d''8 c''8 bes'8 a'8 g'8 |\n  f'2. r4 |\n  d''4. d''8 ees''8 d''8 c''8 bes'8 |\n  c''2 ees''4 g'8 g'8 |\n  f'4 d''4 d''4. c''8 |\n  bes'2. r4\n}\n\nalto = {\n  d'4. d'8 ees'8 d'8 bes8 d'8 |\n  ees'2 ees'4. r8 |\n  ees'4. ees'8 ees'8 ees'8 ees'8 ees'8 |\n  d'2. r4 |\n  d'4. d'8 ees'8 d'8 bes8 d'8 |\n  ees'2 ees'4 cis'8 cis'8 |\n  d'4 f'4 f'4. ees'8 |\n  d'2. r4 |\n  ees'4. ees'8 f'8 ees'8 d'8 c'8 |\n  d'2 f'2 |\n  f'4. f'8 f'8 e'8 e'8 e'8 |\n  f'4 d'4 ees'4 r4 |\n  f'4. f'8 g'8 f'8 ees'8 d'8 |\n  ees'2 g'4 cis'8 cis'8 |\n  d'4 f'4 f'4. ees'8 |\n  d'2. r4\n}\n\ntenor = {\n  bes4. bes8 bes8 bes8 f8 f8 |\n  f4( g4) f4. r8 |\n  a4. c'8 c'8 c'8 a8 a8 |\n  bes2. r4 |\n  bes4. bes8 bes8 bes8 f8 f8 |\n  g2 bes4 bes8 bes8 |\n  bes4 bes4 bes4. a8 |\n  bes2. r4 |\n  a4. a8 bes8 a8 f8 f8 |\n  f2 bes2 |\n  a4. bes8 a8 g8 c'8 bes8 |\n  a4 bes4 c'4 r4 |\n  bes4. bes8 bes8 bes8 f8 f8 |\n  g2 bes4 bes8 bes8 |\n  bes4 bes4 bes4. a8 |\n  bes2. r4\n}\n\nbass = {\n  bes,4. bes,8 bes,8 bes,8 bes,8 bes,8 |\n  c2 f4. r8 |\n  f4. f8 f8 f8 f8 f8 |\n  bes,2. r4 |\n  bes,4. bes,8 bes,8 bes,8 bes,8 bes,8 |\n  ees2 ees4 e8 e8 |\n  f4 f4 f4. f8 |\n  bes,2. r4 |\n  f4. f8 f8 f8 f8 f8 |\n  bes,2 bes,2 |\n  c4. c8 c8 c8 c8 c8 |\n  f2. r4 |\n  bes,4. bes,8 bes,8 bes,8 bes,8 bes,8 |\n  ees2 ees4 e8 e8 |\n  f4 f4 f4. f8 |\n  bes,2. r4\n}\n"}, "harmonizations": [{"slug": "face-to-face-tabernacle-1921", "tune": "face-to-face", "music_file": "face-to-face-tabernacle-1921.ly", "canonical_key": {"tonic": "bes", "mode": "major"}, "time": "4/4", "pickup": null, "system_breaks": [15, 30, 45], "breaks_manual": true, "final_barline": "|.", "amen": false, "verse_variable_melisma": false, "refrain": {"soprano_note_start": 31}, "markings": [{"soprano_note": 31, "kind": "refrain"}], "harmonizer": "grant-colfax-tullar", "arrangement_type": "harmonization", "source_line": "Setting: Grant Colfax Tullar, 1899, as printed in Tabernacle Hymns, No. 2, 1921. Public domain.", "provenance": {"layer": "harmonization", "source": "tabernacle-hymns-no-2-1921", "source_kind": "hand_encoded", "pd_basis": "pub_pre_1931", "pd_evidence": "Four-part setting of Tullar's FACE TO FACE as printed in Tabernacle Hymns, No. 2 (Chicago: Tabernacle Publishing Co., 1921), hymn 62, \"Face to Face.\" \u2014 the leaf carries no printed page number \u2014 headed \"Mrs Frank A. Breck.\" (left) and \"Grant Colfax Tullar.\" (right), marked \"Moderato.\", in B-flat major and 4/4 with no pickup, the word \"Chorus.\" standing at the head of the third system, and footed \"Copyright, 1899 by Tullar-Meredith Co   Used by per\". THAT FOOTER LINE RUNS OFF THE TRIM ON BOTH SCANS and is quoted above exactly as far as it prints; it is not a full transcription of the copyright notice, which the binding cuts. US public domain: the underlying copyright is 1899 and the printing read is 1921, both well before 1931, and no modern arranger stands anywhere in the chain. TWO DATED SCANS OF THAT PLATE WERE READ AND REQUIRED TO AGREE \u2014 archive.org tabernaclehymnsn00chic leaf 0070 (2365 px across, staff space 28.54-28.83 px, median 28.67, i.e. 14.3 px per diatonic step) and archive.org tabernaclehymnsn02chic leaf 0068 (1877 px across, staff space 22.57-22.89 px, median 22.79, i.e. 11.4 px per step); the two items are two scans of the one book, and at this hymn n00chic runs two leaves LATER than n02chic, the same offset the ADELAIDE record notes. THE NOTEHEADS WERE MEASURED FIRST AND THEN READ BY EYE, and it is the eye reading that ships. Per scan, with its own binarisation and its own five-line fits: ink is whatever is darker than 82% of the LOCAL paper level (a 3x3 grey opening, then a 41x41 grey dilation, then a 41x41 box filter), closed 5x5; holes under 0.60 staff space squared, no wider than 1.30 and no taller than 0.80 staff space are filled, which turns each hollow half or dotted-half OUTLINE into a disc without flooding the boxes that staff lines and stems enclose; staff lines are suppressed by vertical run length AFTER that fill; and every head is converted to a diatonic step against a five-line fit INTERPOLATED FROM BAND-BY-BAND MEASUREMENTS around that head's own x, never from a straight fit across the system. Whole-system grid residuals: 0.34-1.28 px mean against a 22.7-28.9 px staff space. Column-table residuals: 0.203 step mean on scan 1 system 1, 0.153 system 2, 0.158 system 3, 0.188 system 4; 0.131, 0.134, 0.137 and 0.191 for the same four systems on scan 2. EVERY ONE OF THE 240 NOTE POSITIONS WAS THEN CONFIRMED BY EYE off grid-overlay crops at 4x to 12x. The eye overruled the machine three times and the second scan, read by eye, agreed with the correction each time: the alto of system 3 note 13 is E4 (machine +1 = F4), the bass chord of system 2 note 11 is B-flat3 over F3 (scan-2 machine +5/+3 = E-flat3 over C3), and the tenor of system 3 note 9 is A3 (scan-2 machine +7 = G3). The whole of SYSTEM 1 was reproduced independently by the scan-2 machine pass \u2014 15 columns, 60 note positions, mean residual 0.131 step \u2014 and matches this encoding in all four voices with 0 disagreements; systems 2, 3 and 4 were read by eye on scan 1 and collated column by column against the scan-2 machine pass, with a scan-2 eye crop taken at every disagreement. ACCIDENTALS WERE SWEPT SEPARATELY AND BY EYE over every column of BOTH leaves; the automated figures above see staff positions only and never covered accidentals. The two-flat signature stands at the head of every staff. The sweep found exactly FOUR printed accidentals and no others: a SHARP before the alto C4 and a NATURAL before the bass E3 at soprano note 24 (system 2, \"Je-\"); the same pair again at soprano note 54 (system 4, \"shall\"); and a NATURAL before the alto E4 at soprano note 42 (system 3, \"star-\"). THE MEASURE RULE WAS THE TRAP AND IT IS APPLIED DELIBERATELY: in systems 2 and 4 the very next chord prints the same G4-over-C4 pair with no second sharp and the same B-flat3-over-E3 pair with no second natural, inside the same bar, so those are C sharp 4 and E natural 3 too. Two independent witnesses on the one plate settle it \u2014 systems 2 and 4 are musically identical here and BOTH print exactly one sharp and one natural followed by one unmarked repetition \u2014 and the harmony confirms it: the chord is E-G-B flat-C sharp, a fully diminished seventh on E (vii dim 7 of the dominant), resolving E to F, C sharp to D and G to F onto the B-flat-over-F that opens the next bar. Had the second chord been C natural over E flat it would have been a C minor seventh, which does not resolve that way. The system-3 natural likewise carries from soprano note 42 to notes 43 and 44 (the alto E4 never moves), giving C7 three times over a C3 pedal into F major; it EXPIRES at the next barline, so the alto E in that last bar is E flat again, which is what makes the chord F7 leading into system 4. EVERY CHORD WAS NAMED. All sixty are diatonic B-flat major or one of three secondary dominants (C7 = V7/V, the E diminished seventh above, and the F7 that is simply V7). Two that surprised the reading and are NOT errors: under \"Sav\" (system 1 bar 2) the plate reads bass C3, tenor F3, alto E-flat4, soprano B-flat4 \u2014 C minor seventh (ii7) with the tenor F standing as a fourth over the bass C and rising to G, which is how the chord completes itself on the second quarter \u2014 and it resolves on \"-iour,\" where the soprano falls to A4 against alto E-flat4, an A4/E-flat4 tritone that is simply the third and seventh of F7. The tenor moves F3-G3 under that suspension, slurred on the plate, making C minor seventh on the second beat. NO SCAFFOLD WAS USED AND NOTHING WAS CHECKED AGAINST ONE: no MIDI, no .nwc, no Cyber Hymnal asset and no existing encoding was opened at any point. No Hymnary incipit cross-check was available either \u2014 Hymnary's /tune/ route returns 403 to this repo \u2014 and none was used. The one file written from this reading was a bare soprano MIDI used for the duplicate guard, which scores 0.58 against its nearest corpus neighbour (EWING), far below the 0.80 threshold. DEPARTURES FROM THE PAGE, in full: (1) TWO FERMATAS ARE NOT ENGRAVED. The plate prints a fermata above the treble staff and an inverted fermata below the bass staff over the last chord of system 3 (soprano note 45, the held \"sky;\") and again over soprano note 53 in system 4 (the held \"-ry,\"). This corpus cannot engrave a fermata, so both are dropped and recorded here. (2) The \"Moderato.\" tempo word is not carried, this corpus holding no tempo marks. (3) AT THE LAST CHORD OF SYSTEMS 2 AND 4 THE PLATE PRINTS FIVE NOTES, not four: the lower staff carries three hollow dotted heads on one down stem, B-flat3 over F3 over B-flat2, under treble B-flat4 and D4. The four voices taken here are the four the voice-leading of the preceding F7 gives \u2014 soprano B-flat4 (from C5), alto D4 (from E-flat4), tenor B-flat3 (the leading tone A3 rising), bass B-flat2 (from F3) \u2014 and the added F3 that fills in the fifth is dropped, because this corpus admits exactly four voices and forbids divisi. That the four voices are right is confirmed by the end of system 1, where the soprano has F4 instead of B-flat4 and the same voice-leading yields a complete triad, and the plate there prints exactly four notes with no added F3. (4) The printed ellipsis after \"sky;\" in the chorus's first lyric line, which marks the held note, is not carried. Nothing else departs: every pitch, every duration, every rest, all four accidentals, the tenor slur in bar 2 and the closing light-heavy double bar are the plate's, and the systems break exactly where the page breaks them.", "pd_jurisdictions": {"US": true}, "encoding_origin": "Hand-encoded from the page images of Tabernacle Hymns, No. 2 (1921), hymn 62 \u2014 archive.org tabernaclehymnsn00chic leaf 0070, collated notehead by notehead against archive.org tabernaclehymnsn02chic leaf 0068. No MIDI, no Cyber Hymnal asset and no existing encoding was used at any point.", "encoding_license": "PD-dedication"}, "music_ly": "%% FACE TO FACE (8.7.8.7. with Refrain) \u2014 Grant Colfax Tullar, 1899. Four SATB\n%% voice streams, canonical key B-flat major, 4/4, NO pickup, sixteen full bars:\n%% four systems of four bars, fifteen soprano noteheads to a system, sixty in all.\n%% The chorus is systems 3-4, i.e. soprano notes 31-60.\n%% RULES (CI-enforced): exactly the variables soprano/alto/tenor/bass/global;\n%% absolute octaves; each voice rhythmically complete; NO chords <>, NO <<>>,\n%% NO \\secondShift/\\unisonWholeShift/tightSpacing, NO \\break, NO \\bar, NO lyrics.\n%%\n%% SOURCE OF THE NOTES. Tabernacle Hymns, No. 2 (Chicago: Tabernacle Publishing\n%% Co., 1921), hymn 62, \"Face to Face.\" \u2014 the leaf carries no printed page\n%% number \u2014 headed \"Mrs Frank A. Breck.\" on the left and \"Grant Colfax Tullar.\"\n%% on the right, marked \"Moderato.\", in B-flat major (two flats) and 4/4 with no\n%% pickup, and footed \"Copyright, 1899 by Tullar-Meredith Co   Used by per\" \u2014\n%% that footer line RUNS OFF THE TRIM on both scans and is quoted here exactly as\n%% far as it prints. Read off TWO independent scans of that same plate:\n%% archive.org tabernaclehymnsn00chic leaf 0070 (2365 px across, staff space\n%% 28.54-28.83 px, median 28.67 = 14.3 px per diatonic step) and archive.org\n%% tabernaclehymnsn02chic leaf 0068 (1877 px across, staff space 22.57-22.89 px,\n%% median 22.79 = 11.4 px per step). The two items are two scans of the one book,\n%% and at this hymn n00chic runs two leaves LATER than n02chic \u2014 exactly as it\n%% does at ADELAIDE, whose record notes the same offset.\n%%\n%% HOW THE NOTEHEADS WERE READ. Per scan, with its own binarisation and its own\n%% five-line fits: ink is whatever is darker than 82% of the LOCAL paper level (a\n%% 3x3 grey OPENING, then a 41x41 grey dilation, then a 41x41 box filter),\n%% followed by a 5x5 binary closing; holes smaller than 0.60 staff space squared\n%% and no wider than 1.30 / taller than 0.80 staff space are filled, which turns\n%% each hollow half or dotted-half OUTLINE into a disc without flooding the boxes\n%% that staff lines and stems enclose; staff lines are suppressed by vertical run\n%% length AFTER that fill; every head is then converted to a diatonic step\n%% against a five-line fit INTERPOLATED FROM BAND-BY-BAND MEASUREMENTS in a\n%% window around that head's own x, never from a straight fit across the system.\n%% Whole-system grid residuals came out 0.34-1.28 px mean against a 22.7-28.9 px\n%% staff space. THE MACHINE PASS IS A FINDER, NOT A DECIDER: every pitch below\n%% was then confirmed BY EYE off grid-overlay crops at 4x to 12x, and it is the\n%% eye reading that ships. It overruled the machine three times \u2014 the alto of\n%% system 3 note 13 (E4, machine +1), the bass chord of system 2 note 11 (B-flat3\n%% over F3, machine +5/+3 on scan 2), and the tenor of system 3 note 9 (A3,\n%% machine +7 on scan 2) \u2014 and in each case the second scan, read by eye, agreed\n%% with the correction.\n%%\n%% THE TWO SCANS AGREE. The whole of system 1 was reproduced independently on\n%% scan 2 by the machine pass (15 columns, 60 note positions, mean residual 0.131\n%% step) and it matches this encoding in all four voices with 0 disagreements.\n%% Systems 2, 3 and 4 were read by eye on scan 1 and collated column by column\n%% against the scan-2 machine pass plus scan-2 eye crops at every disagreement.\n%%\n%% ACCIDENTALS ARE NOT WHAT A NOTEHEAD PROBE SEES. Every column of BOTH leaves\n%% was swept separately BY EYE at high magnification; the figures above cover\n%% staff positions only and never covered accidentals. The two-flat signature\n%% stands at the head of every staff. The sweep found exactly FOUR printed\n%% accidentals in the body of the hymn, and no others anywhere:\n%%   - system 2, soprano note 24 (\"Je-\"): a SHARP in the treble before the alto\n%%     head on the first ledger line below the staff (C sharp 4), and a NATURAL\n%%     in the bass staff before the bass head in the second space (E natural 3);\n%%   - system 4, soprano note 54 (\"shall\"): the same pair again, at the same\n%%     place in the parallel bar.\n%%   - system 3, soprano note 42 (\"star-\"): a NATURAL in the treble before the\n%%     alto head on the bottom line (E natural 4).\n%% THE MEASURE RULE WAS THE TRAP HERE and it is applied deliberately. In systems\n%% 2 and 4 the VERY NEXT chord prints the same G4-over-C4 pair with NO second\n%% sharp and the same B-flat3-over-E3 pair with no second natural; both stand\n%% inside the same bar, so they are C sharp 4 and E natural 3 as well. Two\n%% independent witnesses on the one plate settle it: systems 2 and 4 are\n%% musically identical here and BOTH print exactly one sharp and one natural\n%% followed by one unmarked repetition. The chord is E-G-B-flat-C sharp, a fully\n%% diminished seventh on E \u2014 vii(o)7 of the dominant \u2014 resolving with E to F,\n%% C sharp to D and G to F onto the B-flat-over-F that opens the next bar; had\n%% the second chord been C natural / E flat it would have been a C minor seventh,\n%% which does not resolve that way and which the plate would have had to mark.\n%% The same rule carries the system-3 natural forward from soprano note 42 to\n%% notes 43 and 44 (the alto E4 does not move), giving C7 - C7 - C7 over a C3\n%% pedal into the F major of the next bar; it EXPIRES at that barline, so the\n%% alto E in bar 12 is E flat again, which is what makes that chord the F7 that\n%% carries into system 4.\n%%\n%% NO SCAFFOLD WAS USED AND NOTHING HERE WAS CHECKED AGAINST ONE. No MIDI, no\n%% .nwc, no Cyber Hymnal asset, no Hymnary incipit and no existing encoding was\n%% opened at any point. (Hymnary's /tune/ route is behind a 403 for this repo, so\n%% no incipit cross-check was available; none was used.) The only file written\n%% from this reading was a bare soprano MIDI used for the duplicate guard, which\n%% scores 0.58 against its nearest corpus neighbour (EWING), far below the 0.80\n%% threshold.\n%%\n%% RHYTHM, as the plate prints it. Every system is four bars of 4/4. Bars 1, 5,\n%% 9 and 13 are a dotted quarter and five eighths. Bars 2 and 10 are a half plus\n%% (bar 2) a dotted quarter and an eighth rest, (bar 10) a second half. Bars 6\n%% and 14 are a half, a quarter and two eighths. Bars 3 and 11 are a dotted\n%% quarter and five eighths; bars 7 and 15 are two quarters, a dotted quarter and\n%% an eighth. Bars 4, 8 and 16 are a dotted half and a quarter rest in all four\n%% voices. Bar 12 is the one bar where the outer voices hold and the inner voices\n%% move: soprano and bass sustain a dotted half (each under a printed fermata)\n%% while alto and tenor walk three quarters, all four then resting a quarter.\n%%\n%% TWO PLACES WHERE THE PLATE DOES SOMETHING THE VOICE COUNT MUST ABSORB:\n%%   (1) BAR 2, the tenor alone moves under the held syllable \"Sav-\": it sings\n%%       F3 then G3 as two quarters against the soprano's and alto's half note,\n%%       and the plate draws a SLUR over the pair. That slur is reproduced. The\n%%       chord on the second quarter is C3-G3-E flat 4-B flat 4 = C minor seventh.\n%%   (2) BAR 2 note 8 and the whole of bar 3's bass, the plate prints ONE head\n%%       carrying both an up and a down stem where tenor and bass are in unison.\n%%       Those are written here as the same pitch in both voices, which is what\n%%       the double stem means.\n%%\n%% DEPARTURES FROM THE PAGE, in full \u2014 see the harmonization yaml, which states\n%% them again: (1) the two printed fermatas (over and under the last chord of\n%% system 3, over and under soprano note 53 in system 4) are NOT engraved, this\n%% corpus having no fermata; (2) the \"Moderato.\" tempo word is not carried, this\n%% corpus holding no tempo marks; (3) at the last chord of systems 2 and 4 the\n%% plate's lower staff prints THREE hollow heads, B-flat3 over F3 over B-flat2,\n%% a five-note chord \u2014 the four voices taken here are the four the voice-leading\n%% gives (soprano B-flat4, alto D4, tenor B-flat3 from the leading tone A3, bass\n%% B-flat2 from F3), and the added F3 that fills the fifth is dropped, because\n%% this corpus admits four voices and forbids divisi; (4) the printed ellipsis\n%% after \"sky;\" in the chorus's first lyric line, which marks the held note, is\n%% not carried. Nothing else departs: every pitch, every duration, every rest,\n%% all four accidentals, the tenor slur in bar 2 and the closing light-heavy\n%% double bar are the plate's, and the plate's own four barlines a system are the\n%% barlines here.\n\\version \"2.24.0\"\n\nglobal  = { \\key bes \\major \\time 4/4 }\n\nsoprano = {\n  f'4. f'8 g'8 f'8 d'8 bes'8 |\n  bes'2 a'4. r8 |\n  c''4. bes'8 a'8 g'8 f'8 g'8 |\n  f'2. r4 |\n  f'4. f'8 g'8 f'8 d'8 bes'8 |\n  c''2 g'4 g'8 g'8 |\n  f'4 d''4 d''4. c''8 |\n  bes'2. r4 |\n  c''4. c''8 d''8 c''8 bes'8 a'8 |\n  bes'2 d''2 |\n  c''4. d''8 c''8 bes'8 a'8 g'8 |\n  f'2. r4 |\n  d''4. d''8 ees''8 d''8 c''8 bes'8 |\n  c''2 ees''4 g'8 g'8 |\n  f'4 d''4 d''4. c''8 |\n  bes'2. r4\n}\n\nalto = {\n  d'4. d'8 ees'8 d'8 bes8 d'8 |\n  ees'2 ees'4. r8 |\n  ees'4. ees'8 ees'8 ees'8 ees'8 ees'8 |\n  d'2. r4 |\n  d'4. d'8 ees'8 d'8 bes8 d'8 |\n  ees'2 ees'4 cis'8 cis'8 |\n  d'4 f'4 f'4. ees'8 |\n  d'2. r4 |\n  ees'4. ees'8 f'8 ees'8 d'8 c'8 |\n  d'2 f'2 |\n  f'4. f'8 f'8 e'8 e'8 e'8 |\n  f'4 d'4 ees'4 r4 |\n  f'4. f'8 g'8 f'8 ees'8 d'8 |\n  ees'2 g'4 cis'8 cis'8 |\n  d'4 f'4 f'4. ees'8 |\n  d'2. r4\n}\n\ntenor = {\n  bes4. bes8 bes8 bes8 f8 f8 |\n  f4( g4) f4. r8 |\n  a4. c'8 c'8 c'8 a8 a8 |\n  bes2. r4 |\n  bes4. bes8 bes8 bes8 f8 f8 |\n  g2 bes4 bes8 bes8 |\n  bes4 bes4 bes4. a8 |\n  bes2. r4 |\n  a4. a8 bes8 a8 f8 f8 |\n  f2 bes2 |\n  a4. bes8 a8 g8 c'8 bes8 |\n  a4 bes4 c'4 r4 |\n  bes4. bes8 bes8 bes8 f8 f8 |\n  g2 bes4 bes8 bes8 |\n  bes4 bes4 bes4. a8 |\n  bes2. r4\n}\n\nbass = {\n  bes,4. bes,8 bes,8 bes,8 bes,8 bes,8 |\n  c2 f4. r8 |\n  f4. f8 f8 f8 f8 f8 |\n  bes,2. r4 |\n  bes,4. bes,8 bes,8 bes,8 bes,8 bes,8 |\n  ees2 ees4 e8 e8 |\n  f4 f4 f4. f8 |\n  bes,2. r4 |\n  f4. f8 f8 f8 f8 f8 |\n  bes,2 bes,2 |\n  c4. c8 c8 c8 c8 c8 |\n  f2. r4 |\n  bes,4. bes,8 bes,8 bes,8 bes,8 bes,8 |\n  ees2 ees4 e8 e8 |\n  f4 f4 f4. f8 |\n  bes,2. r4\n}\n", "preferred": true}], "license": "The hymn texts and tunes are in the public domain; per-layer provenance is recorded on each record."}