The NETS translation of the Book of Psalms has been based primarily on the edition of Alfred Rahlfs, Psalmi cum Odis. (SEPTUAGINTA. Vetus Testamentum Graecum Auctoritate Academiae Litterarum Gottingensis editum vol. X.) Goettingen, 1967 [1931]. Though this edition is not a full-fledged critical edition of the Greek Psalter, it is the best available and as such an appropriate text with which to begin.
 At not a few places, Rahlfs enclosed within square brackets items of text which, though they could not, in his judgment, justifiably be regarded as original, nevertheless have widespread support in the textual traditions. Since in all of these cases I agree with Rahlfs' conclusion, I have taken the next step and have excluded these items from NETS without comment.
 Further improvements to Rahlfs' edition have been made in the light of additional textual information (chiefly ii-v CE; especially the famous P. Bodmer XXIV [Rahlfs 2110]) and more recent study. All these, however, have been included in the footnotes to NETS. Nevertheless, there remains good reason to emphasize that a liturgical text such as the Psalter, with its long and intensive transmission history, can hardly be expected to have been fully restored as yet to its pristine purity. I have therefore followed Rahlfs' lead and, in addition to changing his text, have placed within square brackets items whose originality I consider to be suspect. These may be eliminated altogether by a future editor more courageous than I.

General Character
 There can be no doubt that the NETS paradigm of the Greek as an interlinear translation of the Hebrew fits the Book of Psalms like a glove. That is to say, the linguistic relationship of the Greek to the Hebrew is one of dependence and subservience. Yet within that model it has its own profile. Its translation is literal, if literalness is understood to refer to a high degree of consistency in one-to-one equivalence, including not only so-called content words but structural words as well. Thus literalness might be labeled its central characteristic.
 Literalness may, however, be nuanced in several ways. The number of many-to-one Hebrew-Greek equations is relatively high. That is to say, the translator often selected a single Greek word (or root) for several Hebrew counterparts. This feature, semantic leveling, is fully compatible with the interlinear paradigm and for that reason not to be attributed to the translator's lack of literary imagination. Included here are words and concepts that seem to have been his default stock-in-trade. Thus, for example, ANTILEMPTOR ("supporter") and BOETHOS ("helper") translate seven Hebrew words each, and their cognates further underscore the translator's predilection for them. Obviously, these readily came to mind when the translator looked for suitable equivalents to the Hebrew. Similarly, Greek words that convey a relatedness to the "law" (NOMOS) such as "lawlessness" (ANOMIA), "lawless" (ANOMOS), "to act in a lawless manner" (ANOMEO), "to transgress the law" (PARANOMEO), "transgressor of the law" (PARANOMOS) are prominent. Or again a concept like "injustice" (ADIKIA + cognates) receives an emphasis in the Greek disproportionate to the Hebrew. As a result, the reader of NETS will read more about God's "supporting" and "helping" of humans, and similarly more about the Law as yardstick for "good," "evil" and "injustice," than will the reader of NRSV.
 Not surprisingly perhaps the number of one-to-many Hebrew-Greek equations is relatively low. In such cases a single Hebrew word or root is translated by at least two or more Greek words. Some examples of such differentiation are "good" (TWB) rendered by "good" (AGATHOS) and "kind" (CHRESTOS), and "servant/slave" ((BD) translated most often by "slave" (DULOS) but on a few occasions by "child/*servant*/slave" (PAIS). As might be expected, such differentiation of meaning happens with greater frequency on the verbal (more abstract) side of the language than on the nominal (more concrete) side. Hebrew "to turn" (SHWB), for instance, is rendered by twelve different Greek verbs. Nonetheless, though a measure of differentiation occurs, it is typically rather restricted. Thus, in the example just cited, EPI- and APO-STREPHO account for nearly two-thirds of all instances. Again, this phenomenon of limited and restricted differentiation or an overemphasis on word-centered meaning may be said to be at home in an interlinear environment.
 The above is not to preclude that leveling but especially differentiation cannot at times serve purely stylistic purposes, but given the general nature of the Greek translation of Psalms, leveling and differentiation are usually the better explanation.
 An interesting example which could be differentiation for purely stylistic reasons is the equation of Hebrew "to answer" ((NH) with Greek "to listen to" (EPAKUO or EISAKUO), when God is the subject. Though it is not difficult to discern a semantic shift from Hebrew to Greek, it is not certain that one should make a distinction in meaning between the two Greek words. Interestingly, EISAKUO also translates Hebrew "to hear" (SHM(), though EPAKUO does not. Thus in part at least here too a difference in the Greek echoes a difference in the Hebrew, and to that extent it is still expressive of literalness. For NETS I have opted for the synonyms "to listen" (EISAKUO) and "to hearken" (EPAKUO) respectively.
 Other aspects of literalness at both the lexical and grammatical level add further flavor to the Greek of Psalms. As noted, in keeping with the interlinear model our translator had a strong tendency to overemphasize the importance of individual words and formal details of the Hebrew, at the expense of communicating its coherent meaning. For example, in a number of passages he mechanically transferred the Hebrew gender of a pronominal, thereby presenting an exegetical challenge to anyone reading the text independent of the Hebrew: 26:3-4, 73:18, 80:6, 108:27, 117:23, 118:50, 56, 131:6. (I have flagged such instances in the notes.) Indeed, the Greek translator's starting point was at times almost directly opposite to that advocated by modern semanticists. Not infrequently, he began with the individual word and its supposed core meaning rather than with meaning as it is conveyed by the interrelationship of words in syntactic units. Differently put, one could say that he tends to maximize the individual word and to minimize the context in which it stands, rather than vice-versa.
 Though at times the translator might be charged with throwing at his reader the Hebrew text in Greek guise, to call him a hack would be unfair. Instead, as has been suggested, his translating is heavily circumscribed by the interlinear paradigm which informs his task. Indeed, from that perspective it clearly makes little sense to charge him with inadequate knowledge of Greek and lack of stylistic sensitivity. Even in the superscripts where no context comes to his assistance, he at times introduces an interpretive spin. One may note, for example, that although MSHKYL (a type of song) is rather mechanically glossed by "understanding" (SYNESIS), usually the latter is interpretively put into the genitive case (31:1, 51:1, 52:1, 53:1, 54:1, 73:1, 77:1, 87:1, 88:1). That is to say, apparently in the translator's perception the entire Psalm in question has to do with "understanding." Yet, other such interpretive spins in the titles seem less transparent. When, for example, an individual composition is labeled both a psalm and song (in line with the parent text), he usually makes one an attributive of the other. So in Ps 29:1, 47:1, 66:1, 67:1, 74:1, 86:1, 91:1 we have "a psalm of a song" (PSALMOS ODES) and in 65:1, 82:1, 87:1, 107:1 "a song of a psalm" (ODE PSALMU'). Since PSALMOS refers in the first instance to instrumental music and ODE to vocal music, it is not impossible, as has been suggested ad loc., that he had in mind the relative predominance of these two aspects. But what of "praise-song of a song"(?) (AINOS ODES) in 90:1, 92:1, 94:1, since both terms seem to have a vocal reference? Thus it is clear that the translator often puts the form of the text above its meaning. Admittedly he is kinder to his readers in the acrostic Ps 118, when he not only gives the names of the letters of the Hebrew alphabet, which mark the beginning of each alphabetic strophe, but also supplies a standard Greek numerical equivalent.
 Understandably, it is especially idiomatic and figurative language that tends to suffer severely at the hands of a heavily word-based, interlinear translation style. When in 34(35):3 "to empty out the spear" (HRQ TNYT) is translated by "to pour out the sword"—and a big sword at that!—(EKCHEON ROMPHAIAN) we have decidedly unidiomatic Greek. NRSV translates idiomatically "draw the spear" and the Greek translator might have done likewise but did not.
 To be sure, a modern translator, in the nature of the case, tends to overemphasize what is problematic about the work of an ancient translator and underemphasize what causes no problems. In spite of the translation model he used, most of what the Greek translator of Psalms did is intelligible—and that includes many passages in which the Hebrew text is less than clear—, if not idiomatic. He will even at times introduce difference—where the Hebrew text is identical—perhaps for the sake of variation in style, though that is not the only available explanation. An instructive example is Ps 59:7-14 in comparison with Ps 107:7-14.
 Indeed, one can even find some literary sparks, the exceptions that prove the rule of his regular style. So at 48(49):3 where the Hebrew text speaks of "(sons of) man" (BNY )DM) and "(sons of) humans" (BNY )YSH), the translator, rather than resorting to his standard equivalents, decided to render the first phrase by "earthlings" (GEGENEIS) (a word virtually unknown in the Septuagint) and neatly balanced the pair by a "both . . . and" conjunctive. In 10(11):2 the phrase "to shoot in the dark ()PL) at the upright in heart" becomes "to shoot in a moonless night (SKOTOMENE) at the upright in heart," thus employing a picturesque word rare in Greek and unattested elsewhere in the Septuagint corpus. Or again, in 26(27):9 Hebrew "to forsake" (NTSH), elsewhere rendered simply as "to reject" (APOTHEO), is translated by the rare and graphic verb "to throw to the crows" (APOSKORAKIZO), glossed in NETS by "to damn." Ps 109:3 features "morning-star" (HEOSPHOROS) for Hebrew "morning" (SHHR) rather than the less colorful "morning" (ORTHROS) he uses elsewhere. Such literary nuggets are admittedly not many and one would scarcely expect them in a text whose purpose it is to point the reader away from itself, but they do exist and they do add a dimension to our translator's work.

Of Stereotypes, Calques and Isolates
 The vast majority of words in the Psalms have standard Greek meanings, and that includes the so-called stereotypes which due to rigid equation with a single Hebrew counterpart sometimes fit poorly into their context. Among these may be counted PIKRAINO (" to embitter") + cognates, HYPOMONE ("endurance") + cognates , THELO ("to want") + cognates, EIS TELOS ("completely"), STEAR ("fat" meaning "prosperity"), NEPHROS ("kidney" in reference to the seat of the emotions), PSALMOS("psalm"), ANABATHMOS ("step").
 Calques (Greek words with Hebrew meanings) too can be found in Psalms. The most obvious and widespread example is DIATHEKE, a word which no doubt began its life in Jewish usage as a stereotype, but which in the Psalter is clearly a calque and therefore appropriately translated by "covenant" (the dominant sense of Hebrew BRYT, rather than the "testament" of extra-biblical Greek). Other calques in Psalms are ELEOS ("steadfast love"; when it equates with HSD), EULOGEO "to bless" + cognates, PAROIK- ("sojourn"), KYRIOS ("Lord" = YHWH), EIRENE ("peace/*well-being*"), HORIA (pl) ("territory"), ETHNE ("non-Israelite nations"), ADOLESCHEO ("chat/gossip, *ponder*").
 The Greek translator's overemphasis on individual words is most clearly noticeable in his isolate renderings. Typically, in such cases, etymology of the Hebrew word plays a central role. By way of illustration, in Ps 7:7 Hebrew (BRH ("overflow/arrogance/fury") is translated, via (BR = PERAN ("across"), by PERAS ("end")—a word representing entirely different Hebrew lexemes elsewhere in Psalms—and the Greek line as a whole must mean something like: "Be exalted at the ends (deaths?) of my enemies." A glance at Ps 38:4 confirms that PERAS within the Psalter can seemingly refer to the terminus of human life. Or does the line mean what Thomson thought it did: "Exalt Thyself in the borders of mine enemies"? Whatever the case, as a result of the translator's rigid adherence to his interlinear model his text, on the one hand, means something quite different from MT and, on the other, is scarcely intelligible in Greek.

General Approach
 But if the Greek translator's approach to his Hebrew text was all too often indeed what I have alleged it to be, namely, an approach that overemphasizes the importance of the individual word and its representation of the Hebrew, what are the options of a NETS translator to render this "interlinear" into English? Clearly as long as words are fully concordant with their context and the grammar is at least transparent, NETS can be expected to read relatively smoothly.
 Though I have eschewed any rigid policy of one-to-one Greek-English equation, a reasonable effort has been made to reproduce word echoes in the Greek, which may or may not reflect echoes in the Hebrew. In passing it deserves to be mentioned that this effort has not infrequently meant that the reading of the NRSV has been replaced by a synonym in NETS.
 By "reasonable effort" should be understood that the more circumscribed a word's meaning the better it lends itself to representation by a single English equivalent. Not surprisingly, therefore, there are many Greek terms, especially on the nominal (more concrete) side of the language, that have been given a single English equivalent. Thus a word such as OROS ("mountain") has been rendered throughout by "mountain," even when NRSV, for literary reasons, had opted for "hill" to translate the identical Hebrew word (HR). Similarly, HAMARTOLOS (+ cognates) the standard equivalent for both HT) and RSH( has been uniformly rendered "sinner." Countless others have been treated in the same way.
 When I have judged that the Greek translator differentiates vis-à-vis the Hebrew for semantic reasons, I have differentiated in NETS.
 The highest level of one-to-one Greek-English consistency in Psalms will be observable within individual psalms. Whenever feasible word echoes in the Greek have been reproduced in NETS.
 Perhaps the lowest level of one-to-one consistency has been feasible with respect to relational words, notably prepositions, including phrases that function as prepositions. The obvious reason for this is that such words play a primarily structural role and as such carry little fixed meaning. (See further below.)
 I have already touched on the problem of idiomatic and figurative language. Though in replacement type translations idioms and figurative expressions in the source language are typically lost in the process, others are often gained from the target language, thus producing a kind of equilibrium of loss and gain. In Septuagint translationese, however, one would hardly expect that to be the case, and especially in a poetic book such as Psalms we see a decidedly one-way process. In fact, because of its highly literal approach, what is figurative in Hebrew frequently becomes an oddity in Greek. For example, though Hebrew KLYH ("kidney") often figuratively refers to human emotions, NEPHROS in Greek means only "kidney." Therefore any figurative sense is imposed by the (translated) context. Because NETS has opted for a rather literal translation style, what is unidiomatic and nonfigurative in the Greek will often be the same in English.

Selected words and phrases
 The so-called superscriptions or titles to individual Psalms cause a disproportionate measure of difficulty to a translator, modern and ancient alike. The reason for this, in addition to frequent lexical obscurity, is that the words and expressions used are typically without context. Since there is reason to believe that the superscriptions grew in a piecemeal, atomistic fashion, I have treated them in NETS in an atomistic fashion even beyond the NRSV. The reader who is troubled by this may simply ignore the punctuation between discrete items and string them together.
 The Greek translator's word-based approach is as evident in the superscriptions as everywhere else. Yet within those parameters there is some evidence of differentiation, some of which has already been noted. Thus, for example, one encounters both nominatives and genitives without explicit warrant in the Hebrew, and datives as well as EIS + accusative constructions reflecting Hebrew + L. All in all, whether as a result of differentiation or one-to-one equation with Hebrew, there is a variety of words/phrases that may be labeled "expressions of general reference." That is to say, they indicate without much specificity (partly due to lack of context) that x has something to do with y. Where the Greek differs in detail I have followed suit in NETS.
 One of these "expressions of general reference" is TO DAUID (TO ASAPH et al.). Since the Greek translator clearly did not assign authorship per se to such Psalms, I have opted for the reasonably neutral phrase "Pertaining to Dauid" (et al.), since it allows for a range of perceived connections with the person(s) in question.
 NRSV's "Psalm" has for reasons of tradition been retained and has even been left with initial upper case. It is by no means certain, however, that Greek PSALMOS (psalm), at the time of the Greek translator, was already a technical term. As noted earlier, I have regarded it as stereotype, which means that it still had its chiefly instrumental rather than vocal sense and as such referred in the first instance to a tune being played on a stringed instrument (harp or lyre), which was then plucked (PSALLO) with the fingers rather than struck with a plectrum. Along the same lines, DIAPSALMA, which consistently renders Hebrew SLH ("Selah"), a word of uncertain meaning, has been interpreted to mean not simply an interlude but an interlude on strings and rendered accordingly in NETS. Evidently in recognition of the fact that it indicated a pause in the musical proceedings, the Greek translator did not represent "Selah" when it stood at the close of a psalm (3:9, 23:10, 45:12).
 Another frequent item in the titles is ODE, the regular translation of SHYR ("song"). Though "song" would have been defensible for NETS, I have opted, with some hesitation, for "Ode" instead. To be sure, ODE is the standard equivalent of SHYR (36x) and and may be a calque since it already occurs in the Greek Pentateuch as an equivalent for SHYR, on five occasions within Psalms (32:3, 39:4, 95:1, 97:1, 149:1) the translator opted for AISMA ("song"). This may suggest that he meant to differentiate the two. The presence of KAINON ("new") with all five instances of AISMA may further indicate that an ODE was thought of as being more of a traditional song in distinction from a new and ad hoc composition. But the latter point can clearly not be pressed in light of ODEN KAINEN ("new ode") in 143:9. Consequently, whatever the precise difference in the Greek, I have sought to mirror a difference in NETS by translating the terms as "Ode" and "song" respectively.
 Since the Greek Psalter provides no evidence that the translator made any attempt at distinguishing between the divine names YHWH ("Yahweh," including the short form "Yah") and )DNY ("Adonai"), I have in accordance with NETS policy rendered all occurrences of KYRIOS when representing either by "Lord."
 Hades, one of very few translations into Greek which has not been translated into English, is probably a calque. If that is correct, to translate it simply by English "hell" would seem to be as inappropriate as rendering its Hebrew counterpart, Sheol, by that term.
 There is good reason to believe that the translator's chief reason for using SKENOMA, and by extension KATASKENOO, is to reserve SKENE for the Tabernacle, in line with Pentateuchal usage. Consequently, it appears unlikely that SKENOMA+ is meant to carry any special nuance, apart from its general sense of non-permanence. In Psalms, therefore, both noun and verb have been rendered by "tent," in distinction from "Tabernacle," which renders SKENE (sg). (The fact that English "to tent" is not common usage appropriately reflects the relative uncommonness of the verb and noun in Greek usage.) Since in a number of passages the notion of non-permanence seems to stand in tension with the context, it is possible to regard the root partly as a stereotype. It should be noted, however, that Hebrew SHKN is sometimes translated by a word of permanence.
 Greek DULOS in Psalms has been rendered, with some hesitation, by "slave." The hesitation stems purely from the connotation "slave" carries in modern English. The Greek Pentateuch translated Hebrew (BD ("servant"/"slave") almost exclusively by Greek PAIS, a word with a very similar semantic range. Lev 25:44 and 26:13 where DULOS is found instead, confirm the difference between the two Greek terms. The translator of Psalms chose almost exclusively DULOS ("slave") (53x) with only four occurrences of PAIS. Greek Isaiah and Jeremiah present a mix with Ezekiel inclining toward DULOS and the Minor Prophets completely in the latter camp. There is insufficient evidence to argue that DULOS in Psalms is a calque.
 Since ANABATHMOS has no prior history within the LXX in the sense of "ascents" it can scarcely be called a calque. Since both inside and outside the LXX, the word is used only in the sense of steps (of a stairway), it evidently carries its normal semantic range. What we have, therefore, is a stereotype, which arose because the translator insisted on the equation (LH ANABAINO > HM(LWT > ANABATHMOI. Why he did not choose ANABASIS instead—which occurs in 83:6 but for a different Hebrew root—is perhaps surprising, but suggests that he was not thinking of "ascending/ascension" as such. Accordingly, NETS readers will read Songs of the Steps rather than Songs of Ascents. Of interest is that according to Jewish exegetical tradition these psalms were recited by the Levites on the fifteen steps leading from the court of the men to that of the women in the temple.

Some Problems of Grammar
 Since the Psalms are poetic literature and furthermore typically do not tell a story with a timeline of its own, the Hebrew verbal categories of tense-aspect presented somewhat of a problem to the Greek translator. Quite clearly he ended up by working with a set of default or unmarked equations. Thus the Hebrew prefixed conjugation is normally translated by the Greek future indicative, the affixed conjugation by the aorist indicative and the participle often by the present indicative. It should be noted, however, that participles are also often translated by participles even though that may produce either strained syntax in the Greek or an obvious change in syntax.
 Though he had his default equivalents, apparent awareness of contextual sense brought about some flexibility. Nevertheless, the defaults produced a certain stiltedness and abruptness in the use and sequence of Greek tenses. In an effort to communicate at least some of this quality to the reader of NETS, I have routinely rendered all Greek aorists by the English simple past even when the NRSV has used the present perfect—which in NETS Psalms has been reserved for the Greek present perfect.
 A second grammatical item that occasioned some discomfort is the preposition. The primary function of prepositions is grammatical, that is to say, they forge relationships between constituents of sentences. Nonetheless, like full words or lexemes but unlike lexically empty structural markers such as case endings, prepositions have some lexical content. They may appropriately be referred to as (semantically) bleached words or lexemes in distinction from (semantically) full words or lexemes. Once this is realized, it comes as no surprise that when prepositions are treated as though they were full lexemes with a set meaning, idiomatic usage tends to go out the window (cf. the earlier note on idiomatic language). Yet this is precisely what often happens in the Greek Psalms due to the fact that its translator tended to read his Hebrew text in an atomistic manner. The result is that each Hebrew preposition was typically given a default equivalent in Greek, and this default tended to be used irrespective of whether it produced idiomatic Greek. Since the issue here is one of infelicity in Greek style (which is difficult to mimic in English) rather than change in meaning from the Hebrew, such instances have been ignored for the purposes of NETS, lest one produce senseless English! The rule of thumb has been to ignore all instances of default equations between Hebrew and Greek, but to take seriously instances of non-default equations, the assumption being that when the translator deviated from his routine he did so for a reason.
 The same rule of thumb has been applied to conjunctive KAI ("and"), the standard equivalent of Hebrew W ("and"). When renderings other than KAI were selected, I have tried to signal this in NETS if the NRSV had not already made any change unnecessary. Since, however, for reasons of English style the NRSV often ignores W, and occasionally adds an "and" without explicit warrant in the Hebrew, I have followed a second rule of thumb: NETS has been made to deviate from the NRSV only when the Hebrew and Greek texts are in disagreement. The only scenario not covered by this rule is the following: MTvsNRSV=LXX=NETS, since MT and LXX show disagreement while NRSV and NETS show agreement. For whatever reason this configuration seldom occurs in Psalms.
 The question of articulation has been deemed largely a matter of Greek style and not easily transmitted to English. On occasion, however, I have introduced a discrepancy between NRSV and NETS, when such was thought to be demanded by the Greek text.

Gender Specific Language
 The gender-inclusive language of the NRSV has been changed only when it seems demanded by the Greek text. Though Hebrew (YSH, (DM and GBR are regularly translated by ANTHROPOS ("human being") from time to time they are rendered instead by gender-specific ANER ("man/male"). As a result, where the latter appears in Greek NETS has been made gender-specific, on the assumption that the Greek translator made a choice. The same has been done with Greek HUIOS ("son") versus TEKNON ("child"), both representing Hebrew BN.

 Since punctuation in Rahlfs' edition of the Greek Psalter is largely based on his personal editorial decision, I have, for the purposes of NETS, not always felt bound by it, nor have deviations from it been entered in the notes.
 Unlike "Selah" in the NRSV, its equivalent in NETS, "interlude on strings", is consistently followed by an empty line on the grounds that any interlude signals a pause.
 In their division of materials the Masoretic and the Greek Psalters do not completely agree. Since MT Psalms 9 and 10 are a single psalm in the Greek (9) and Greek Psalm 113 is two psalms in MT (114, 115), between these points the numeration of LXX is up by one from MT. Further, since LXX Psalms 114 and 115 equal MT 116:1-9 and 116:10-19 respectively, and MT 147:1-11 and 12-20 constitutes LXX 146 and 147, between these points we have an identical number of Psalms, though their enumeration varies by one.
 Versification in MT, LXX and NRSV is potentially more confusing. While MT and LXX coincide, the NRSV excludes the psalm titles from the numbering and as a result is out of step with both MT and LXX. NETS follows the Rahlfs numbering for chapters and verses but gives the NRSV numbers in parentheses.

 Apart from standard grammars and lexica, I have benefited from other English translations of the Greek Psalter such as those of Brenton (1844) and Thomson (1808) (see the NETS To The Reader), but particularly from the more recent renditions by A. Lazarus, The Holy Psalter from the Septuagint (Madras, 1966), by the Fathers of the Holy Transfiguration Monastery, The Psalter according to the Seventy (Boston, 1987), and by José M. de Vinck and Leonidas C. Contos, The Psalms Translated from the Greek Septuagint (Allendale NJ, 1993). A copy of the The Psalter according to the Seventy was kindly sent to me by Father Basil. It should be borne in mind, however, that the aim of NETS has been distinctly different from any of these. Among the more exegetical treatments Martin Flashar's "Exegetische Studien zum Septuagintapsalter" (ZAW 32 [1912] 81-116, 161-98, 241-68) has been invaluable, as has been F. W. Mozley, The Psalter of the Church (Cambridge, 1905).
 My recent graduate students in Septuagint at the University of Toronto, Cameron Boyd-Taylor, Paul McLean, Tony Michael, Marc Saunders, and Tyler Williams, have been of great help. A special debt of gratitude I owe, however, to Cameron for his incisive and persistent critique which has immeasurably improved the final product.

Albert Pietersma
August 1998.