INTRODUCTION
The Byzantine Text
The Byzantine text is the historically dominant form of the Greek New Testament. As a result, it was the Textus Receptus, a close relative of the Byzantine text compiled from a small number of manuscripts, that was the dominant form of the printed Greek New Testament from the early sixteenth century to the late nineteenth century. In 1881, however, the Textus Receptus was effectively supplanted by Westcott and Hort's Greek New Testament, particularly in academic circles. Westcott and Hort prepared their Greek text on the assumption that there was a recension of the Byzantine text in the fourth century that became the basis for all subsequent Byzantine manuscripts. Based on this assumption, Westcott and Hort counted (or discounted) the overwhelming majority of Byzantine manuscripts as originating from a single formal recension source, removing them from the equation, so that they could give preference to a small handful of manuscripts, particularly Codex Vaticanus (B) and Codex Sinaiticus (ℵ). Although the assumption of a fourth century recension has now largely been discredited due to a complete lack of evidence, Westcott and Hort's preference for a small handful of manuscripts has endured, and the modern critical editions of Nestle-Aland and UBS have become the standard Greek text accepted in academic circles today.
Yet there are critical flaws in the underlying methodology of the reasoned eclecticism that is practiced in the editions of Nestle-Aland and UBS. In his essay “The Case for Byzantine Priority,” Dr. Maurice Robinson makes the following observation:
Modern eclecticism creates a text which, within repeated short sequences, rapidly degenerates into one possessing no support among manuscript, versional, or patristic witnesses. The problem deteriorates further as the scope of sequential variation increases.
In other words, when the text-critical decisions of the editors of Nestle-Aland and UBS are considered over the course of a few verses (and sometimes over the course of only one verse), it is often the case that the resulting text as a whole has no support in any Greek manuscript, ancient translation, or quotation from the church fathers; rather, it is a conjectural text. This critical flaw of the modern eclectic approach has never been adequately addressed by its proponents. For this reason and others, some prefer the Byzantine text, which is based on the overwhelming majority of Greek manuscripts.
The Byzantine text is not quite the same as the Textus Receptus, which is the textual basis of the New Testament in the King James Version and the New King James Version. While the Textus Receptus is within the Byzantine family of texts, the first edition of Erasmus' Greek New Testament was produced from only seven manuscripts. Although those manuscripts were from the Byzantine family, they contained some readings that have very little support among Greek manuscripts.
On average,* Here the word average refers to the median rather than the mean. when there are variants among Greek manuscripts, the readings adopted by Robinson and Pierpont are supported by 96% of the Greek manuscripts in the Gospels,† Except for the Pericope Adulterae (John 7:53–8:11), in which the Greek manuscripts are fairly evenly divided between three main families. 90% of the Greek manuscripts in Acts and the Epistles, and 64% of the Greek manuscripts in Revelation. These Byzantine manuscripts, which number in the low thousands, represent many individual streams of transmission. And while they are generally later in date, they were all copied from earlier manuscripts of the same text type. Even Westcott and Hort acknowledge that the Byzantine text dates at least as far back as the fourth century, which is contemporaneous with Codex Vaticanus (B) and Codex Sinaiticus (ℵ). Thus the Byzantine textform is ancient, highly uniform, and well attested by a variety of independent streams of transmission. Therefore it has a strong claim toward being the original text of the New Testament. Those seeking further information are encouraged to read Robinson's essay in full.
Colophons
Many Greek manuscripts include interesting scribal notes in the colophons of the Gospels and Pauline epistles. In the Gospels these notes give the date of publication. In the Pauline epistles they give details about the place of authorship, who delivered the epistle, and, in the pastoral epistles, details about the recipient. Because these are scribal notes and not the sacred text itself, they should not be considered infallible. However, most readers do not even realize that these notes exist, especially in the Gospels. Consequently, the colophons are included as footnotes in this edition to allow readers to easily access and evaluate them.
Editions of the Greek New Testament Compared in this Volume
The Text-Critical Greek New Testament is an edition of Robinson and Pierpont's 2018 Greek text‡ The following minor modifications have been made to Robinson and Pierpont's text: movable nu and movable sigma have been removed when they occur before a consonant, paragraph breaks have been modified, poetic formatting has been added, certain accent marks around clitics have been modified (without changing the meaning of any words), and the various forms of Θεος, Χριστος, Κυριος, and Πνευμα Αγιον have been capitalized (except when they do not refer to the Godhead). However, apart from these superficial modifications, the actual text and punctuation have not been altered. that documents every difference found in the following editions and manuscript families of the Greek New Testament.§ Differences between movable nu and movable sigma are ignored. In the footnotes movable nu and movable sigma are removed when they occur before a consonant. Differences between meaningless word breaks are also ignored (see Appendix E). When such words are written in the footnotes, the spelling of Robinson and Pierpont is followed. For the purposes of comparison, typographical errors in the compared editions have been corrected. See Appendix B for a list of corrections.
| ANT | Greek New Testament of the Ecumenical Patriarchate of Constantinople, edited by Basileios Antoniades (1904) |
| BYZ | Robinson and Pierpont's Alternate Byzantine Readings (2018)* Since 2018 Robinson revised his electronic text for the alternate Byzantine readings in four verses (Philem. 1:1; Rev. 3:2; 7:5; 9:10). These revised alternate readings are listed below. Philem. 1:1 χριστου ιησου ¦ ιησου χριστου Rev. 3:2 ἔμελλες ἀποβάλλειν ¦ ἤμελλες ἀποβάλλειν ¦ ἔμελλον ἀποθανεῖν Rev. 7:5 ἐσφραγισμέναι ¦ ἐσφραγισμένοι Rev. 9:10 ἐξουσίαν ἔχουσιν τοῦ ¦ καὶ ἡ ἐξουσία αὐτῶν |
| CT | Critical Text (This designation is used when NA, SBL, TH, and WH are all in agreement. In Mark, Acts, and the Catholic Epistles, this designation is used when ECM, NA, SBL, TH, and WH are all in agreement) |
| ECM | Editio Critica Maior for Mark, Acts, and the Catholic Epistles (1997-2022) |
| ECM* | This designation is used to mark the variants that appear in ECM on a split guiding line.† A split guiding line lists two or more variants as the mainline text. The use of a split guiding line indicates that the ECM editors have left open the decision as to which of the variants they believe might be the initial text. When three variants appear on a split guiding line, all three variants are listed in the footnotes of this edition, even if one agrees with the main Byzantine text. This happens in only four verses (Acts 9:43; 13:46; 17:3; 21:13). When two variants appear on a split guiding line, only the variant or variants that disagree with the main Byzantine text are listed. |
| HF | Hodges and Farstad, The Greek New Testament According to the Majority Text, 2nd edition (1985) |
| NA | Nestle-Aland (This designation is used when NA27 and NA28 are in agreement.) |
| NA27 | Nestle-Aland, 27th edition (1993) |
| NA28 | Nestle-Aland, 28th edition (2012) |
| PCK | Wilbur Pickering, The Greek New Testament According to Family 35, 3rd edition (2020) |
| SBL | SBL Greek New Testament (2010) |
| SCR | Scrivener's Textus Receptus (1894) |
| ST | Stephanus' Textus Receptus, 3rd edition (1550) |
| TH | The Greek New Testament, Produced at Tyndale House, Cambridge (2017) |
| TR | Textus Receptus (This designation is used when SCR and ST are in agreement.) |
| WH | Westcott-Hort (1881) |
In addition to the editions listed above, the following manuscript families are documented for the Pericope Adulterae (John 7:53–8:11).‡ See Appendix C for more information about manuscript families in the Pericope Adulterae.
| μ1 | A very late family of approximately 12 manuscripts |
| μ2 | An early family of approximately 47 manuscripts that is slightly favored by NA and WH |
| μ3 | A somewhat early family of approximately 37 manuscripts |
| μ4 | An early family of approximately 36 manuscripts that has four long omissions |
| μ5 | An early family of approximately 298 manuscripts that is followed by Robinson and Pierpont |
| μ6 | A somewhat early family of approximately 204 manuscripts that is followed by Hodges and Farstad |
| μ7 | A late family of approximately 261 mansucripts that is followed by Pickering and slightly favored by TR |
| λ1 | A somewhat early family of approximately 224 lectionary manuscripts of John 8:3–11 that is followed by ANT in 8:3b–11 |
| λ2 | An early family of approximately 40 lectionary manuscripts of John 8:1–11 that is followed by ANT in 8:1b–8:3a |
| pt | A marker indicating that a family is split between two or more variants |
The following manuscript families are documented for the book of Revelation.§ See Appendix D for more information about these manuscript families. These families are documented only when there is a general consensus* For the purposes of this volume, K is considered to have a general consensus when Hodges and Farstad's apparatus shows that a reading is supported by Ma (but not Mapt). Similarly, Αν is considered to have a general consensus when Hodges and Farstad's apparatus shows that a reading is supported by Md and Me (but not Mdpt or Mept). Hodges and Farstad's families Md and Me correspond with Hoskier's Egyptian and Erasmian families, respectively, and together provide a good representation of Αν. In Revelation 7:5–8, however, Hodges and Farstad's apparatus incorrectly indicates that the TR readings are supported by Md and Me. For those notes, Hodges and Farstad's apparatus is not followed. In a few other instances, where Hodges and Farstad are silent, K and Αν have been determined directly from Hoskier. for the family and the family differs from the Robinson and Pierpont text.
| K | The main Koine tradition in Revelation comprised of approximately 78 disparate manuscripts that represent many copying eras and locations |
| Αν | A family of approximately 58 manuscripts in Revelation that contain or derive from the fourth-century commentary of Andreas of Caesarea |
The Textus Receptus
While it is common to refer to the Textus Receptus as a single entity, in reality there are various editions of the Textus Receptus, which all differ from one another. Although Erasmus was the first to publish what became known as the Textus Receptus, it was Robert Estienne (Stephanus) who came to shape the text as we know it today. Stephanus' third edition (published in 1550 and known as Editio Regia or the “Royal Edition”) is a splendid masterpiece of typographical skill. It was also the first printed edition of the Greek New Testament to include text-critical notes in the margins. Modifying Stephanus' text, Theodore Beza published five editions of the Textus Receptus. His fifth edition (published in 1598) was one of the primary source texts of the Greek New Testament used by the translators of the King James Version. At times, however, the King James Version deviates from Beza's fifth edition. Seeking to recreate the Greek text underlying the New Testament translation of the King James Version, Scrivener modified Beza's fifth edition with readings from various editions of the Textus Receptus that the King James translators would have had at their disposal. Scrivener published his modification of Beza's fifth edition in 1881. When people think about the Textus Receptus today, they think primarily of Stephanus' 1550 edition and Scrivener's 1881 edition.
Editions of the Critical Text
Westcott and Hort published their Greek New Testament in 1881, basing their text-critical decisions on the possibility that a majority of manuscripts could descend from a single formal recension source and thus should not necessarily be preferred as correct. Although they never proved this possibility from the actual manuscript evidence, their theory paved the way for future editions of the critical text. Following in the footsteps of Westcott and Hort, the Nestle-Aland editions have become the standard Greek text in most academic circles today. Closely aligned with the Nestle-Aland editions is the Editio Critica Maior, which thus far has only published Mark, Acts, and the Catholic Epistles. The Editio Critica Maior is unique in the sense that it uses a split guiding line for hundreds of readings. This means that, in many instances, there is no single base text. When compared to the twenty-seventh edition of Nestle-Aland, the changes introduced in the Editio Critica Maior at times move in the direction of the Byzantine Text. Another modern critical text that presents slightly different readings is the SBL Greek New Testament, edited by Michael Holmes. Following the same general methodology as the editors of Nestle-Aland, Holmes differs from Nestle-Aland in over six hundred places, providing an alternate perspective within the eclectic tradition. A fourth critical text that presents slightly different readings is The Greek New Testament, Produced at Tyndale House, Cambridge, which its editors say is rooted in the earliest manuscripts and relies upon the study of scribal habits to inform text-critical decisions.
Modern Editions of the Byzantine Text
Although the Byzantine text is quite stable for the vast majority of the New Testament, in the Pericope Adulterae (John 7:53–8:11) and the book of Revelation the degree of variation among Byzantine manuscripts increases significantly. Partly in response to this high degree of variation in the Pericope Adulterae and the book of Revelation, Wilbur Pickering published The Greek New Testament according to Family 35. Family 35 (also known as Kr) is a large family of highly uniform manuscripts within the Byzantine text tradition. It is the only family of manuscripts that has a demonstrable archetype for every book of the New Testament. This means that even in the Pericope Adulterae and the book of Revelation, there is little question as to the reading of Family 35. Many, however, argue that the high level of uniformity among manuscripts in Family 35 is the result of a systematic recension. Whatever the case may be, the readings of Family 35 at times represent fewer than 20% of extant Greek manuscripts, and there are no extant manuscripts for this family prior to the eleventh century. Nevertheless, Pickering's edition provides important documentation of a large but late family within the Byzantine text tradition.
In addition to the Textus Receptus and Family 35, the present volume also documents variants found in The Greek New Testament According to the Majority Text, edited by Zane Hodges and Arthur Farstad. The edition of Hodges and Farstad differs very little from that of Robinson and Pierpont with the exception of the Pericope Adulterae and the book of Revelation, where it follows a stemmatic approach for determining the original Greek text. Using this stemmatic approach, Hodges and Farstad hypothesize family trees to show the relationships of various manuscript families. They then make text-critical decisions based on those hypothetical family trees. This approach provides an alternate perspective to the main Byzantine textform.
The Greek New Testament of the Ecumenical Patriarchate of Constantinople, edited by Basileios Antoniades, provides one further witness to the Byzantine text family. This edition relies more heavily on readings found in Greek lectionaries than any other edition of the Greek New Testament. At times it includes readings with very little support among Greek manuscripts. Many of these readings are printed in small type in the 1904 and 1912 editions to indicate doubt on the part of the editor as to their originality. This text, also known as the Patriarchal Text, is used in the Greek-speaking Orthodox Churches.
Robinson and Pierpont's Alternate Byzantine Readings
In addition to documenting the variants found in the editions described above, The Text-Critical Greek New Testament also documents Robinson and Pierpont's alternate Byzantine readings. For the bulk of the New Testament, Robinson and Pierpont follow Von Soden's family Kx. When Kx is nearly evenly divided, Robinson and Pierpont generally follow the portion of Kx that is also supported by Kr, while listing the alternate Byzantine reading in the margin.
In the Pericope Adulterae (John 7:53–8:11), the Byzantine manuscript tradition is nearly evenly divided between three main families known as μ5, μ6, and μ7 (which is closely linked to Kr). Robinson and Pierpont follow μ5, Hodges and Farstad follow μ6, and Pickering follows μ7. When μ5 is nearly evenly divided, Robinson and Pierpont list the alternate μ5 readings in the margin. They also list in the margin the primary readings of μ6 as well as the alternate readings of μ6 when that family is nearly evenly divided.
In Revelation, there are three large families of manuscripts. K represents the main Koine tradition in Revelation and is comprised of approximately seventy-eight disparate manuscripts that represent many copying eras and locations.† Family K is also known as Q or 𝔐K. Αν is comprised of approximately fifty-eight manuscripts that contain or derive from the fourth-century commentary of Andreas of Caesarea.‡ Family Αν is also known as 𝔐A. This family is much less cohesive than K, frequently being divided in support of two or more readings. The third family is the Complutensian group. It is comprised of approximately thirty-two manuscripts that are highly uniform and tend to align with the readings of the Complutensian Polyglot. This family is closely linked to Kr and generally agrees with either K or Αν. These three families account for nearly 60% of the manuscripts of Revelation.
As is the case with the Pericope Adulterae, editors of the Byzantine text take different approaches to the three main manuscript families in Revelation. The Textus Receptus often follows Αν, but this is due more to an accident of history than to any intentional decision on the part of the various editors. Pickering follows the Complutensian family exactly. Hodges and Farstad follow K very closely, departing from it only on rare occasions. Robinson and Pierpont also generally prefer the readings of K. At times, however, they follow Αν, particularly when a significant number of K manuscripts abandon their group consensus and align with the Αν reading.§ In a few instances, Robinson and Pierpont depart from K due to other transmissional and orthographic considerations. Whenever a variant unit is nearly evenly divided among two or more readings, Robinson and Pierpont list the alternate reading(s) in the margin.
Text-Critical Footnotes
For the purpose of simplicity, the text-critical footnotes of this volume generally ignore punctuation, capitalization, accents, and breathing marks. However, capitalization, accents, and breathing marks are written in the footnotes when necessary to differentiate meaning. Text-critical signs such as brackets, diamonds, double angle brackets, and small type are also generally ignored because of the level of complexity this would add to the footnotes. Nevertheless, the use of double brackets is at times documented in the text-critical notes, particularly when used by ECM and/or NA.
Manuscript Percentages
For sets of variants that have been fully collated in the Text und Textwert volumes, the manuscript percentages supporting the exact Greek text for each variant are listed.* In the book of Philemon, the percentages for variants not collated in Text und Textwert have been calculated from Matthew Solomon's collation. In the book of Jude, the percentages for variants not collated in Text und Textwert have been calculated from Joey McCollum's tabulations of Tommy Wasserman's collation. In John 18, percentages have been calculated from Michael Morrill's collation. In John 7:53–8:11, percentages have been calculated from a draft of Maurice Robinson's collation dated January 23, 2024. It should be noted that, while manuscript percentages are not the sole factor to be considered in the task of textual criticism, they should not be ignored either, particularly when they demonstrate the dominance of a particular text type. (See Appendix A for details about the calculation of manuscript percentages.)
An analysis of the Text und Textwert data yields the manuscript percentage averages listed in the tables below. The RP percentages are based on every variant unit presented in Text und Textwert.† The Text und Textwert volumes present a total of 1,043 variant units. However, the collations for five of those units are incorrect. Those five variant units are therefore excluded from the percentage of manuscript calculations. In 166 variant units all the editions compared in this volume agree. (See Appendix A for more information.) Solomon's and Wasserman's collations are not considered in the calculation of these averages. The percentages for all the other editions apply only when the editions differ from the RP text. Due to the presence of outliers in the data, the median is presented along with the mean, as the median may very well provide a truer picture of the “average” manuscript percentages. Using the tables below, the reader can make a general estimate of the percentage of manuscripts supporting any given reading that is not documented in Text und Textwert. However, the reader should be aware that any given variant may deviate greatly from the averages presented below.
Gospels
| Edition | Mean | Median |
| RP | 91.9% | 95.8% |
| BYZ | 35.4% | 37.0% |
| PCK | 28.6% | 28.6% |
| ST | 18.5% | 16.2% |
| SCR | 18.3% | 16.4% |
| TR | 17.7% | 15.3% |
| ANT | 13.7% | 7.7% |
| TH | 3.2% | 1.0% |
| CT | 2.9% | 1.0% |
| SBL | 2.9% | 1.0% |
| WH | 2.8% | 1.0% |
| NA27 | 2.7% | 1.0% |
| HF | — | —‡ Text und Textwert does not document any of the HF variants in the Gospels. |
Acts & Epistles
| Edition | Mean | Median |
| RP | 86.1% | 89.7% |
| BYZ | 32.5% | 38.8% |
| HF | 32.3% | 40.6% |
| PCK | 28.7% | 25.9% |
| ANT | 16.1% | 13.6% |
| TR | 15.7% | 8.1% |
| ST | 15.2% | 8.1% |
| SCR | 14.6% | 9.2% |
| TH | 6.6% | 4.4% |
| CT | 6.5% | 4.4% |
| SBL | 6.3% | 4.3% |
| NA27 | 6.2% | 4.3% |
| WH | 6.0% | 4.1% |
Revelation
| Edition | Mean | Median |
| RP | 64.4% | 63.7% |
| PCK | 36.8% | 38.2% |
| BYZ | 36.7% | 37.4% |
| HF | 35.7% | 36.8% |
| ANT | 24.5% | 25.6% |
| TR | 24.1% | 22.9% |
| SCR | 23.8% | 22.9% |
| ST | 23.7% | 22.9% |
| CT | 16.5% | 11.3% |
| WH | 16.4% | 11.3% |
| SBL | 16.1% | 10.1% |
| TH | 15.8% | 10.0% |
| NA27 | 15.6% | 9.5% |
For the Pericope Adulterae (John 7:53–8:11), manuscript percentages have been calculated from Robinson's collation.§ SBL and TH do not include the Pericope Adulterae and are therefore exluded from the list of averages.
Pericope Adulterae
| Edition | Mean | Median |
| RP | 55.7% | 52.2% |
| PCK | 67.7% | 67.2% |
| HF | 61.3% | 64.0% |
| ST | 60.0% | 64.7% |
| SCR | 59.9% | 64.7% |
| TR | 59.9% | 64.7% |
| ANT | 59.8% | 64.7% |
| NA | 57.4% | 65.8% |
| WH | 55.9% | 64.7% |
| BYZ | 42.1% | 37.9% |
*^ Here the word average refers to the median rather than the mean.
†^ Except for the Pericope Adulterae (John 7:53–8:11), in which the Greek manuscripts are fairly evenly divided between three main families.
‡^ The following minor modifications have been made to Robinson and Pierpont's text: movable nu and movable sigma have been removed when they occur before a consonant, paragraph breaks have been modified, poetic formatting has been added, certain accent marks around clitics have been modified (without changing the meaning of any words), and the various forms of Θεος, Χριστος, Κυριος, and Πνευμα Αγιον have been capitalized (except when they do not refer to the Godhead). However, apart from these superficial modifications, the actual text and punctuation have not been altered.
§^ Differences between movable nu and movable sigma are ignored. In the footnotes movable nu and movable sigma are removed when they occur before a consonant. Differences between meaningless word breaks are also ignored (see Appendix E). When such words are written in the footnotes, the spelling of Robinson and Pierpont is followed. For the purposes of comparison, typographical errors in the compared editions have been corrected. See Appendix B for a list of corrections.
*^ Since 2018 Robinson revised his electronic text for the alternate Byzantine readings in four verses (Philem. 1:1; Rev. 3:2; 7:5; 9:10). These revised alternate readings are listed below. Philem. 1:1 χριστου ιησου ¦ ιησου χριστουRev. 3:2 ἔμελλες ἀποβάλλειν ¦ ἤμελλες ἀποβάλλειν ¦ ἔμελλον ἀποθανεῖνRev. 7:5 ἐσφραγισμέναι ¦ ἐσφραγισμένοιRev. 9:10 ἐξουσίαν ἔχουσιν τοῦ ¦ καὶ ἡ ἐξουσία αὐτῶν
†^ A split guiding line lists two or more variants as the mainline text. The use of a split guiding line indicates that the ECM editors have left open the decision as to which of the variants they believe might be the initial text. When three variants appear on a split guiding line, all three variants are listed in the footnotes of this edition, even if one agrees with the main Byzantine text. This happens in only four verses (Acts 9:43; 13:46; 17:3; 21:13). When two variants appear on a split guiding line, only the variant or variants that disagree with the main Byzantine text are listed.
‡^ See Appendix C for more information about manuscript families in the Pericope Adulterae.
§^ See Appendix D for more information about these manuscript families.
*^ For the purposes of this volume, K is considered to have a general consensus when Hodges and Farstad's apparatus shows that a reading is supported by Ma (but not Mapt). Similarly, Αν is considered to have a general consensus when Hodges and Farstad's apparatus shows that a reading is supported by Md and Me (but not Mdpt or Mept). Hodges and Farstad's families Md and Me correspond with Hoskier's Egyptian and Erasmian families, respectively, and together provide a good representation of Αν. In Revelation 7:5–8, however, Hodges and Farstad's apparatus incorrectly indicates that the TR readings are supported by Md and Me. For those notes, Hodges and Farstad's apparatus is not followed. In a few other instances, where Hodges and Farstad are silent, K and Αν have been determined directly from Hoskier.
†^ Family K is also known as Q or 𝔐K.
‡^ Family Αν is also known as 𝔐A.
§^ In a few instances, Robinson and Pierpont depart from K due to other transmissional and orthographic considerations.
*^ In the book of Philemon, the percentages for variants not collated in Text und Textwert have been calculated from Matthew Solomon's collation. In the book of Jude, the percentages for variants not collated in Text und Textwert have been calculated from Joey McCollum's tabulations of Tommy Wasserman's collation. In John 18, percentages have been calculated from Michael Morrill's collation. In John 7:53–8:11, percentages have been calculated from a draft of Maurice Robinson's collation dated January 23, 2024.
†^ The Text und Textwert volumes present a total of 1,043 variant units. However, the collations for five of those units are incorrect. Those five variant units are therefore excluded from the percentage of manuscript calculations. In 166 variant units all the editions compared in this volume agree. (See Appendix A for more information.) Solomon's and Wasserman's collations are not considered in the calculation of these averages.
‡^ Text und Textwert does not document any of the HF variants in the Gospels.
§^ SBL and TH do not include the Pericope Adulterae and are therefore exluded from the list of averages.