© 2014 by David B. Sloan, Ph.D.
The majority view among New Testament scholars over the last century has been the two-source or four-source hypothesis, with Matthew and Luke making independent use of both Mark and a now-lost source, Q. This paper will present the case for the two-source hypothesis, first by looking at the differences between Matthew, Mark, and Luke in the triple tradition material (passages present in all three of these gospels); then by looking at differences between Matthew and Luke in the double tradition material (passages present only in Matthew and Luke); and finally by considering the double tradition passages by themselves. These three approaches correspond roughly with the arguments for Markan priority, for Matthean and Lukan independence of one another, and for Q as a single document.
In considering the triple tradition we can make three observations that support Markan priority: (1) the brevity of Mark, (2) the primitivity of Mark, and (3) divergences in order and wording.
Streeter begins with the observation that ancient writers incorporated earlier works into their own. Noting that most of the pericopes in Mark have a parallel in Matthew and Luke, he argues that it is more likely that the briefer text of Mark was incorporated into Matthew and Luke, with the omission of just a few difficult or less important passages than that Mark abridged Matthew and Luke, omitting the birth narratives, most of Jesus' teachings, and the resurrection appearances. That Mark is not abbreviating Matthew and/or Luke is clear from the fact that
where the two Gospels are parallel, it is usually Matthew, and not Mark, who does the abbreviation. For example, the number of words employed by Mark to tell the stories of the Gadarene Demoniac, Jairus' Daughter, and the Feeding of the Five Thousand are respectively 325, 374 and 235; Matthew contrives to tell them in 136, 135 and 157 words. Now there is nothing antecedently improbable in the idea that for certain purposes an abbreviated version of the Gospel might be desired; but only a lunatic would leave out Matthew's account of the Infancy, the Sermon on the Mount, and practically all the parables, in order to get room for purely verbal expansion of what was retained. On the other hand, if we suppose Mark to be the older document, the verbal compression and omission of minor detail seen in the parallels in Matthew has an obvious purpose, in that it gives more room for the introduction of a mass of highly important teaching material not found in Mark.
Second, Mark's wording of the stories is more primitive than Matthew's and Luke's. This can be seen in a number of ways. First, Matthew and Luke use better grammar than Mark. Sometimes Matthew and Luke agree in the better expression; more often they each have a different better expression than Mark's, suggesting not only that Matthew and Luke are improving Mark, but that they are doing so independently. Second, Mark has more Aramaic expressions than Matthew and Luke. Third, Matthew and Luke speak more reverentially of Jesus than Mark. So in Mark Jesus is addressed as κύριε ("Lord") only once, whereas in Matthew and Luke we find this 19 and 16 times respectively. Likewise whereas Mark 3:5 says Jesus "was not able to do any miracles" in Nazareth, Matt 13:58 says Jesus "did not do many miracles there." It is easier to believe that Matthew reworded the Markan expression to not detract from Jesus' power than that Mark reworded the Matthean expression. Finally, Mark's primitivity can be seen in that what is in Mark and not in Matthew and Luke is often a redundancy. Streeter notes, "Mark reads like a shorthand account of a story by an impromptu speaker - with all the repetitions, redundancies, and digressions which are characteristic of living speech." It is hard to believe that these emerged from Mark rewording Matthew's or Luke's text, but very easy to see that the absence of redundant expressions in Matthew and Luke is due to Matthew and Luke using Mark as a source. Sometimes when redundancies are absent in Matthew and Luke they agree on which portion of Mark to retain; other times they do not. This suggests again that Matthew and Luke use Mark independently of one another.
Third, Matthew and Luke often agree with Mark in order and wording, but when one differs from Mark, the other gospel typically agrees with Mark. In fact, Matthew and Luke never agree against Mark in the placement of a pericope, and when Matthew and Luke share a saying or pericope that is not in Mark they never place it in the same context, except for in the case of John's preaching and the temptation narrative, which could not possibly be put in a different position than alongside where they are in Mark. This suggests that Mark is the middle term and that if Matthew and Luke are following Mark, they are doing so independently of one another. In other words, one of the following scenarios must be the case:
Our consideration of the triple tradition has given three reasons for accepting Markan priority. Within these three we have seen reasons for thinking that Matthew and Luke are independent of one another. Three more reasons can be given for Matthean and Lukan independence when we consider their handling of the double tradition material: (1) alternating primitivity; (2) omissions; and (3) different placement.
Whereas Mark consistently gives a more primitive reading than Matthew and Luke, when we consider the double tradition material we find that sometimes Matthew gives the more primitive reading and sometimes Luke does. For example, in the Beelzebul controversy Matthew has "by the Spirit of God" (Matt 12:28) where Luke has "by the finger of God" (Luke 11:20). With Luke's interest in the Holy Spirit is unlikely that he would remove the word from Matthew for a "more obscure" wording, and so Luke has the more original wording. Similar examples can be found in Luke 6:20b-23; 11:2-4, 30, 49-51; 12:8-9, 46. Other times Matthew has the more primitive reading (Matt 5:45-47; 7:24-27; 10:34; 23:23; etc.). Sometimes Matthew and Luke are each more primitive than the other in the same passage. For example, in the temptation narrative Matthew expands the quotation of Deuteronomy in the bread temptation and changes "Jerusalem" to "the holy city" in the temple-pinnacle temptation, while Luke adds an explanation of how Satan can give authority in the idolatry temptation. Or in the Lord's prayer, the shorter reading in Luke is more primitive, but Matthew's "debts" is replaced with "sins" by Luke, though Luke does not replace "those who are indebted to us" (Matt 6:12; 11:4). Or in the saying, "Away from me all workers of lawlessness," Luke agrees exactly with Psalm 6:9 on the first four words (ἀπόστητε ἀπʼ ἐμοῦ πάντες) but diverges on the last four (where Luke has ἐργάται ἀδικίας), whereas Matthew agrees exactly with the psalm on the last four words (οἱ ἐργαζόμενοι τὴν ἀνομίαν) but diverges on the first four (where Matthew has ἀποχωρεῖτε ἀπʼ ἐμοῦ). Matthew and Luke must be editing a source that followed Psalm 6:9 exactly, and one has preserved half of the quotation while the other preserved the other half.
Second, what Matthew and Luke each omit from the other's material would be inexplicable if one had access to the other. Matthew often makes additions to Markan pericopes, but Luke omits them (Matt 3:15; 12:5-7; 13:14-17; 14:28-31; 16:16-19; 19:9, 19b; 27:19, 24). This is especially difficult to understand in regard to Matthew's scriptural fulfillments (Matt 1:22-23; 2:15; 2:17-18; 2:23; 4:13-16; 8:17; 12:15-21; 13:35; 21:4-5; 26:56; 27:9-10; cf. 2:5-6), considering that Luke never omits a scripture quotation from Mark. Luke also has an interest in Peter, so "it seems doubtful that Matt 16.17-19 would have made no impact at all on Luke." Other additions Matthew makes to Markan pericopes can be found elsewhere in Luke (mostly in his travel narrative, Luke 9:51-18:14), but always in a different context, suggesting that Luke had access not to Matthew but to Matthew's source. But it is not only insertions Matthew makes into Mark that Luke does not follow. Matthew and Luke each omit the other's infancy narrative and resurrection appearances. With Luke's interest in depicting the Herods negatively, why would he omit Matthew's details on Herod from the infancy narrative if he knew it? And there are countless M passages that are congenial to Luke's purposes that are omitted (e.g., the sheep and the goats in Matt 25:31-46). Finally, Burkett notes that "[n]ot only passages, but also less noticeable features of Matthew's style such as recurring words, phrases, grammatical constructions, themes, and redactional techniques characteristic of Matthew are absent from both Mark and Luke," such as Matthew's use of προσκυνέω with Jesus (Matt 2:2; 2:8; 2:11; 8:2; 9:18; 14:33; 15:25; 20:20; 28:9; 28:17), a concept that is important to Luke, making Luke's omission of it rather telling. The same can be said going the other direction. Matthew omits Luke's infancy narrative, his call narrative with its focus on Peter, and his resurrection appearances. Matthew seems to be aware of Luke's source, but not of Luke.
Finally, we note how Matthew and Luke never agree on the placement of double tradition material, except for John's preaching and the temptation narrative, which only have one logical place in the Markan frame. Matthew regularly places the double tradition material either into a Markan narrative or into large blocks of teaching that he has added to his Markan frame. Luke places most of the double tradition material in blocks of non-Markan narratives that are found between Markan blocks. Burkett notes:
Matthew places the "Sermon on the Mount" between Mark 1:21 and 1:22, while Luke places his corresponding "Sermon on the Plain" later between Mark 3:21 and 3:22. Matthew has the material "Jesus and John" [a chapter after the Sermon on the Mount,] between Mark 2:22 and 2:23, while Luke has it [with the Sermon on the Plain] between Mark 3:21 and 3:22.
And it is not just that they choose to place it differently. They seem to be unaware of the other's placement. Kloppenborg notes:
Luke betrays no awareness of the particular ways that Matthew attached these to Mark's framework. Luke's woes against the Pharisees (11.37-54) are delivered in the house of an anonymous Pharisee who lived outside of Judea. Matthew's woes, delivered to a crowd, are attached to Mark's attack on the scribes (12.37b-40), posed out-of-doors in Jerusalem. . . . And in the case of the Jerusalem word (13.34-5), Luke has also eschewed its Matthean setting (in Jerusalem), placing it in some nameless location in Samaria in response to the comments of more extra-Judaean Pharisees (13.31-3). . . .
Similarly, Luke's apocalyptic materials in Luke 12.39-59; 17.23-37; 19.11-27 betray no influence of the fact that Matthew has them spoken to the disciples on the Mount of Olives. Instead, Luke 17.23-37 is framed as speech in Samaria (17.11-19), where again Pharisees are present (17.20) (!). Luke 12.39-59 belongs to discourses delivered to growing crowds somewhere in . . . Galilee or Samaria (12.1). Luke 19.11-27 is also moved backward to a point where Jesus is between Jericho and Bethany (rather than in Jerusalem) and spoken in response to the comments of some anonymous interlocutors. In short, Luke has not only consistently altered the geographical settings that Matthew gave to this material, but he has also consistently changed the addressees. If Luke knew Matthew, he has curiously not retained any of the framing elements of Matthew's version.
This is especially surprising when we consider how careful Luke is to follow Mark's order. Why would Luke be so careful with one source and work so haphazardly with the other?
The same holds true if we suppose that Matthew was aware of Luke's gospel. Whereas Luke has Jesus preach the Sermon on the Plain to his disciples immediately after the call of the disciples taken from Mark 3:16-19, Matthew has Jesus' disciples come to him for the Sermon (Matt 5:1) before he includes the call of the disciples (Matt 10:1), so that the reference to "his disciples" in Matt 5:1 is the first use of μαθηταί in the book. Likewise Matthew seems unaware of the fact that Luke has placed Jesus' response to John after stories of Jesus healing the sick and raising the dead. Nor does Matthew seem to be aware of the fact that Luke makes that connection clear by stating (in clearly Lukan language) that "in that hour he healed many from diseases and plagues and evil spirits and granted sight to many blind people" (Luke 7:21). Matthew thus loses the great context Luke gives for Jesus' saying.
It thus becomes clear that Matthew and Luke both had independent access to Mark and to a source or sources for the double tradition material. The question remains where this double tradition material comes from. Three bits of evidence reveal the great likelihood that Q is a single, written document: (1) the presence of doublets in Matthew and Luke; (2) agreements in order; and (3) the coherence of Q.
That Matthew and Luke are working from two sources is evidenced by the presence of doublets. For example, Mark 8:34 ("If anyone wants to follow me, let him deny himself and take up his cross and follow me") is copied almost identically in Matt 16:24 and Luke 9:23, but then Matt 10:38 and Luke 14:27 have a slightly different form of the same saying. This suggests that they have a second source and have both preserved the forms they found in each of their sources. The same can be seen with sayings in Mark 4:25; 8:34, 35, 38; 9:37; 13:12, 31; 14:21. Sometimes the presence of a doublet causes either Matthew or Luke to not include either the Markan version or the Q version (Mark 3:28-30; 4:21, 22; 6:7-13; 8:11-13; 9:40; 10:11-12, 31; 11:22-23, 24; 12:38-39; 13:11, 21, 35). Still other times we will find Matthew and Luke both following the Q version and not the Markan version (Mark 1:2, 7-8; 3:22-27; 4:24cd, 30-32; 9:42, 50a). In each of these cases except for the preaching of John the Baptist, Luke has nothing in the position where the Markan equivalent would go. He entirely skips over it as if he has already planned on using the Q equivalent later in his gospel.
Despite the fact that Matthew and Luke cannot agree where in their gospels to place the double tradition material, they agree on what order the material goes in.So Luke's Sermon on the Plain has the same material in the same order as Matthew's Sermon on the Mount, even though they place this material in a different place in Mark. And both of them follow the Sermon on the Mount with the story of the centurion asking that his servant might be healed. Likewise, Jesus' response to John's disciples is followed by a teaching about John, which is followed by a word against "this generation" (Matt 11:2-19; Luke 7:18-35). Taylor has shown that in each of the Matthean speeches the double tradition material is found mostly in the order in which it appears in Luke. Thus it appears that Matthew has gone through a document from beginning to end looking for things to add into his lengthy speeches. If Matthew and Luke were independently using loose traditions, the order would not agree. If Luke were using Matthew it is not clear why Luke would break up Matthew's speeches, distributing the contents relatively evenly between Luke 10 and Luke 18. It seems that they have a shared document before them.
If we remove everything from Luke that is not double tradition material, what remains is a coherent document with a logical outline; a distinct style, vocabulary, and theology; and distinct motifs. So Kloppenborg notes:
the elements that appear to be key in the organization of Q are not the key elements of Matthean (or Lukan) redaction: Q is not programmatically interested in Jesus as Torah-observant and as a fulfilment of Torah, nor as a messianic shepherd, nor in Lukan themes such as Jesus' piety, meal settings, reconciliation, or euergetism. Conversely, the Lot cycle and deuteronomistic theology are not the key organizing elements of either Matthew or Luke. The logical point is that the double tradition (along with a few Mark-Q overlaps) exhibits a thematic coherence that does not derive from Matthew's [or Luke's] redactional interests.
We see within the double tradition a distinct style. Like Mark, and unlike Matthew and Luke, Q begins many sentences with καί. As with the triple tradition material, some of these uses of καί are replaced with δέ in Matthew, and some are replaced in Luke, but the evidence that their source used an introductory καί is clear. The different style reveals that neither of the authors of Matthew and Luke is the original author of the double tradition material. The same can be said of vocabulary. Fleddermann notes that many of the Matthew-Luke agreements use words or expressions that are found nowhere in Matthew or Luke-Acts outside of Q passages (Matt 4:5/Luke 4:9; Matt 7:3-5/Luke 6:41-42; Matt 7:5/Luke 6:42; Matt 8:8/Luke 7:6; Matt 11:11/Luke 7:28; Matt 8:20/Luke 9:58; Matt 6:11/Luke 11:3; Matt 12:25/Luke 11:17; Matt 12:28/Luke 11:20; Matt 6:22-23/Luke 11:34, 36; Matt 10:28/Luke 12:4; Matt 10:32/Luke 12:8). Catchpole notes that there is "some sort of space between the theology of Q and the theology of Matthew . . . on key issues such as poverty, the Gentiles, the debate about the status of the law, the kingship and Davidic lineage of Jesus, the relationship of Jesus to Wisdom, and the gift of the Spirit." This is not to say that Q and Matthew are at odds with one another (obviously Matthew incorporates these elements from Q), but that their presentation in regard to these matters is distinguishable. In fact, sometimes we can see in Luke how Matthew has changed a wording to tone down Q emphases, and we see the same in Matthew with the Lukan changes. Tuckett notes that Q has a prevalent wisdom Christology, which "is neither Matthean nor Lukan." Kloppenborg argues that the introduction to Q contains an allusion to the Sodom and Gomorrah story that neither Matthew nor Luke picks up, though the allusion is confirmed by the repetition of the Lot motif in numerous subsequent passages in the double tradition. Parker notes that "the style of M does not pervade Q and the style of Q does not pervade M." All of this supports Harnack's claim that "as soon as one calls to mind the content of the three gospels and compares Q with it, then Q appears to be undoubtedly more homogeneous than any of the three." The coherence of Q strongly suggests that we have not a series of unrelated traditions, but a single, unified document that Matthew and Luke use alongside Mark as their other key source.
Our study of the triple tradition led us to conclude that Matthew and Luke both made use of Mark in compiling their gospels. This was clear based on the brevity of Mark, the primitivity of Mark's wording, and the divergences in order and wording of the triple tradition material. Our study of Matthew's and Luke's handling of the double tradition led us to conclude that Matthew and Luke worked independently from one another and therefore must have both had access to the same sources. This was clear based on alternating primitivity in Matthew and Luke, omissions of Matthean and Lukan material, and different placement of the double tradition material. Our study of the double tradition material led us to conclude that the shared source was a single document, with its own order of material, coherence, distinct style, distinct vocabulary, distinct theology, and distinct motifs. These observations have led the majority of biblical scholars to accept the two-document or four-document hypothesis. To this day there has not arisen a hypothesis that better explains the data.
Catchpole, David R. The Quest for Q. London: T&T Clark, 1993.
Downing, F. Gerald. "Redaction Criticism: Josephus's Antiquities and the Synoptic Gospels. Part 2: Luke and the Other Two Synoptists." Pages 180-199 in New Testament Interpretation and Methods. Edited by Stanley E. Porter and Craig A. Evans. The Biblical Seminar 45. Sheffield: Sheffield Academic Press, 1997. Repr. from Journal for the Study of the New Testament 9 (1980): 29-48
Harnack, Adolf von. The Sayings of Jesus: The Second Source of St. Matthew and St. Luke. New Testament Studies 2. New York: G. P. Putnam's Sons, 1908.
Hawkins, John C. "Probabilities as to the So-Called Double Tradition of St. Matthew and St. Luke." Pages 91-138 in Studies in the Synoptic Problem, ed. W. Sanday. Oxford: Clarendon, 1911.
Fleddermann, Harry T. Q: A Reconstruction and Commentary. Biblical Tools and Studies. Leuven: Peeters, 2005.
Kloppenborg, John S. "City and Wasteland: Narrative World and the Beginning of the Sayings Gospel (Q)." Pages 145-160 in How Gospels Began, ed. Dennis E. Smith. Semeia 52. Atlanta: Scholars, 1991.
__________. "On Dispensing with Q? Goodacre on the Relation of Luke to Matthew." New Testament Studies 49 (2003): 210-236.
Kloppenborg Verbin, John S. Excavating Q: The History and Setting of the Sayings Gospel. Minneapolis: Fortress, 2000.
Parker, Pierson. The Gospel before Mark. Chicago: University of Chicago Press, 1953.
Streeter, Burnett Hillman. The Four Gospels: A Study of Origins. London: Macmillan, 1924.
Taylor, Vincent. "The Original Order of Q." New Testament Essays (1959): 95-118.
Tuckett, Christopher M. Q and the History of Early Christianity: Studies on Q. London: T&T Clark, 1996.
Vassiliadis, Petros. ΛΟΓΟΙ ΙΗΣΟΥ: Studies in Q. Atlanta: Scholars, 1999.
 Burnett Hillman Streeter, The Four Gospels: A Study of Origins (London: Macmillan, 1924), 155-157.
 Streeter, Four Gospels, 158.
 Streeter, Four Gospels, 163.
 Streeter, Four Gospels, 161.
 John C. Hawkins, "Probabilities as to the So-Called Double Tradition of St. Matthew and St. Luke," in Studies in the Synoptic Problem (ed. W. Sanday; Oxford: Clarendon, 1911), 91-138, esp. 102; Streeter, Four Gospels, 268; Petros Vassiliadis, ΛΟΓΟΙ ΙΗΣΟΥ: Studies in Q (Atlanta: Scholars, 1999), 29; John S. Kloppenborg Verbin, Excavating Q: The History and Setting of the Sayings Gospel (Minneapolis: Fortress, 2000), 20; Delbert Burkett, Rethinking the Gospel Sources: Volume 2: The Unity and Plurality of Q (Atlanta: Society of Biblical Literature, 2009), 11.
 Streeter, Four Gospels, 183; Harry T. Fleddermann, Q: A Reconstruction and Commentary (Biblical Tools and Studies; Leuven: Peeters, 2005), 60-65.
 John S. Kloppenborg, "On Dispensing with Q? Goodacre on the Relation of Luke to Matthew." New Testament Studies 49 (2003): 210-236, esp. pp. 224-225; Burkett, Unity and Plurality, 21.
 Kloppenborg, "On Dispensing," 224; cf. David R. Catchpole, The Quest for Q (London: T&T Clark, 1993), 1-59; Burkett, Unity and Plurality, 21. Burkett notes on the last example, "While Luke shows no aversion to the term 'hypocrites' (Luke 6:42; 12:56; 13:15), Matthew shows a positive affinity for it, using it fourteen times. It is more likely therefore that Matthew introduced the term than that Luke dispensed with it." Likewise, with regard to Luke 11:30, Burkett notes: "It is easy to understand why Matthew would have added a reference to Jesus' resurrection, but less easy to understand why Luke would have removed it had he known it" (Unity and Plurality, 21).
 Fleddermann, Q, 61-62.
 Helmut Koester, Ancient Christian Gospels: Their History and Development (London: SCM Press, 1990), 131.
 Kloppenborg, "On Dispensing," 221-222; Vassiliadis, ΛΟΓΟΙ ΙΗΣΟΥ, 29. Kloppenborg notes that "it is manifest from Luke's own treatment of the passion narrative that he is keen to shift the blame from the Romans to the high priests. Pilate's wife's dream (Matt 27.19, inserting into Mark 15.10-11) and Matthew's hand-washing scene (27.24, added to Mark 15.15) would have served Luke's purposes admirably, especially since the dream declares Jesus to be δίκαιος, which is precisely what Luke's centurion says of Jesus (23.47), and the handwashing scene has Pilate declare Jesus to be innocent, something that Luke has redactionally added to Mark at Luke 23.4, 14, 22." ("On Dispensing," 222)
 Catchpole, Quest, 53.
 Kloppenborg, "On Dispensing," 222.
 Vassiliadis, ΛΟΓΟΙ ΙΗΣΟΥ, 29.
 Kloppenborg, "On Dispensing," 223.
 Kloppenborg, "On Dispensing," 223. Kloppenborg also mentions Matt 18:23-35 and 20:1-16 as passages that would have reinforced Luke's purposes rather well.
 Burkett, Unity and Plurality, 27-28.
 Burkett, Unity and Plurality, 13; see also his helpful table on pp. 11-13.
 Kloppenborg, "On Dispensing," 233. Streeter (Four Gospels, 183) is often quoted in this regard:
If then Luke derived this material from Matthew, he must have gone through both Matthew and Mark so as to discriminate with meticulous precision between Marcan and non-Marcan material; he must then have proceeded with the utmost care to tear every little piece of non-Marcan material he desired to use from the context of Mark in which it appeared in Matthew - in spite of the fact that contexts in Matthew are always exceedingly appropriate - in order to re-insert it into a different context of Mark having no special appropriateness. A theory which would make an author capable of such a proceeding would only be tenable if, on other grounds, we had reason to believe he was a crank.
 Christopher M. Tuckett, Q and the History of Early Christianity: Studies on Q (London: T&T Clark, 1996), 9.
 All the examples here are taken from the lengthier discussion in Fleddermann, Q, 54-60. See also Vassiliadis, ΛΟΓΟΙ ΙΗΣΟΥ, 37; Burkett, Unity and Plurality, 24-26.
 Vincent Taylor, "The Original Order of Q." New Testament Essays (1959): 95-118; Vassiliadis, ΛΟΓΟΙ ΙΗΣΟΥ, 37-38; Fleddermann, Q, 46-54.
 Tuckett, Q, 35-36.
 Catchpole, Quest, 5.
 Taylor, "Original Order."
 Fleddermann, Q, 65-68; Vassiliadis, ΛΟΓΟΙ ΙΗΣΟΥ, 38; Kloppenborg, "On Dispensing," 225.
 Kloppenborg, "On Dispensing," 225.
 Fleddermann, Q, 44-45.
 Catchpole, Quest, 6-7.
 Tuckett, Q, 38.
 John S. Kloppenborg, "City and Wasteland: Narrative World and the Beginning of the Sayings Gospel (Q)," in How Gospels Began (ed. Dennis E. Smith; Semeia 52; Atlanta: Scholars, 1991), 145-160, esp. pp. 149-152.
 Pierson Parker, The Gospel before Mark (Chicago: University of Chicago Press, 1953), 30; quoted in Vassiliadis, ΛΟΓΟΙ ΙΗΣΟΥ, 19.
 Adolf von Harnack, The Sayings of Jesus: The Second Source of St. Matthew and St. Luke (New Testament Studies 2; New York: G. P. Putnam's Sons, 1908), 167.