T of C 
Qin as
Qin in
/ Song
Analysis History Ideo-
Personal email me search me
LQXS   /   Da Hujia   /   Xiao Hujia   /   18 Songs     /   Original lyrics 網站目錄
10. Eighteen Blasts of the Nomad Flute 1
- Fugu tuning/mode: 1 3 5 6 1 2 3 3
Hujia Shibapai
My transcription, page 14      
There are a number of qin melodies on the theme of hujia (a term difficult to translate). All retell a version of the story of Cai Wenji abducted by Central Asian nomads. All use the same tuning and are generally played as purely instrumental melodies, and most are attributed to the famous Tang dynasty qin player Dong Tinglan. Today the best known of these seems to be the Hujia Shibapai instrumental melody apparently dating from 1689 and still played.5 However, there are other versions with different melodies, some related, some completely unrelated.

Only one version is clearly designed for singing, the one introduced on the present page. Although also called Hujia Shibapai, its music is unrelated to that of any of the purely instrumental melodies of this name. Instead it is a long qin song first published in Luqi Xinsheng (1597).6 This 1597 version, republished in 1611, is sometimes attributed to the 1597 handbook's own author or compiler, Xu Shiqi, but the basis for this is unclear.7 The lyrics, a first person narration, are the ones attributed to Cai Wenji herself.8 Some of the later versions of Hujia Shibapai do mention or quote these Cai Wenji lyrics, but they are all melodically unrelated to the song published in 1597 and 1611.9

The earliest surviving melodies that have in their title the name Hu Jia10 (nomad reed flute) are Xiao Hujia and Da Hujia. My interpretation of these is included in my Shen Qi Mi Pu CD set. One of the publications of Da Hujia had lyrics applied to it, but it is not really a qin song - the melody is the same as that of the 1425 version, with lyrics added for uncertain reasons.11

Although the 1597 Hujia Shibapai is the earliest known version of this story set to lyrics as an actual qin song, it is possible that there were others at that time.12 A number of opera scripts with related titles have survived.13

According to my understanding this melody was reconstructed in the 1950s or 1960s both by Zha Fuxi and by Chen Changlin; it is not clear whether they consulted each other's work.14 Since then the basic melody has clearly had quite some popularity, but these seem to be based on transcriptions by Wang Di, which do not identify who actually did the reconstruction.15 As for any current popularity, this originally seems to have been due to orchestral and/or opera performances that were based on the Wang Di transcriptions.16

More recently there have also been qin versions that follow the basic transcription in the vocal part but not in the qin part itself. The qin melody is simplified, presumably with the aim of avoiding the difficulties caused by the word-intensive nature of the traditional method for pairing qin and voice.17

The continuing interest in taking this old melody and presenting it to a larger audience should be further encouraged by the recent publication of what seems to be almost the complete work of Chen Changlin, both transcriptions and recordings. Thus the 2013 Chen Changlin Guqin Collected Works (8 CDS) finally makes available a complete recording of the guqin melody (though without the singing); an accompanying publication has Chen's own transcription.18

Some years ago I wrote out my own transcription of the 1597 version of Hujia Shibapai, but have not worked on it with a singer, so I consider my note values quite tentative.19 I did the transcription without consulting any other ones, in particular the one by Wang Di in her Qin Ge. Her transcription is said to come from 1611, but that edition is missing the first page; otherwise the tablature seems identical to that of 1597.


Original Preface
None here

Melody and Lyrics
Eighteen sections; the setting is largely syllabic

Translated in Women Writers, pp. 23 - 29.

Footnotes (Shorthand references are explained on a separate page)

1. "18 Songs" vs. "18 Blasts"
The "pai" of "Hujia Shiba Pai" generally means "rhythm" or "beats". For the scroll it is translated as "songs" because there are 18 verses or songs accompanying the scroll. For the melody it is here translated "blasts", suggesting the music of the "nomad flute" (hujia, which might also be translated as "barbarian reed pipe"). Since the present 1589 version has lyrics, perhaps it would be better translated at "18 Songs of the Nomad Flute"; however, the later melodies called Hujia Shiba Pai generally have no lyrics.

2. 18 Blasts of the Nomad Flute (胡笳十八拍 Hujia Shibapai)
30073.359 胡笳十八拍 tells the basic story then mentions various related poetic and musical texts. This title is more commonly associated with a Qing dynasty instrumental melody of this title.

3. Fugu tuning/mode (復古調 fugu diao)
For Fugu tuning lower the first string and raise the 5th string each one step. This tuning is also called Wuyi Diao, Huangzhong Diao, and other names. For more on mode see Modality in early Ming qin tablature

As for the specific modal characteristics here (based mainly on analyizing the last notes of phrases), the tonal center often shifts between la (with mi as the secondary tone) and do (in which case so becomes the secondary tone).

A challenging aspect is intonation, in particular determining whether the source and/or intention of the melody (from or intended for an opera or play?) affected intonation in the qin melody. At that time (ca. 1600) the way of writing qin tablature was in the process of changing from an old "less precise" system to the modern decimal system (comparative chart). The latter has more potential for indicating small changes in intonation (slight sharpening or flattening of pitches), but the system in this Hujia Shibapai seems to combine the two methods (suggesting it is transitional between the two) and it does not seem to use them at all consistently. Such charts, in fact, indicate finger positions based on mathematical calculations of where they should be on the ideal qin, but the number of inconsistencies here suggest that perhaps the transcriber may have simply been looking at where the left hand fingers seemed to stop the string on the particular instrument that was being played - a notoriously unreliable method .

A case in point is the note "do": when played on the third string (open string pitch so; see on this sample chart) it is most commonly indicated as played at "九八" (9.8), which in theory should make it slightly sharp, but in many places the standard position "十" (10) is indicated instead, and when played up one octave, where the standard decimal position would be 5.6, it is usually indicated as 五八 (5.8), suggesting perhaps it was played slightly flat. In addition, sometimes this higher position is indicated at "五半" (5 1/2), which in theory could suggest a sharp do, but there is a passage in Section 12 where this note is played together with an open 5th string, meaning it is almost certainly intended to be do natural.

In other cases the question arises about to what extent the reconstructer should adjust finger positions based on mode as expressed in other melodies dating from that time, or whether the possible connection of this melody with the opera idiom at the time brought in different non-pentatonic notes or otherwise unknown intonations.

Perhaps further intensive research might yield answers to this problem. Meanwhile such issues have been a further factor in my own delay in reconstructing this quite appealing melody. I suspect they have also been a factor leading to the orchestral versions that seem simply to have ignored the modal and intonational aspects of the original text.

4. Image: page 1 of my transcription
See pdf copy.

5. Hujia Shibapai as played today
See under Da Hujia for information on recordings and transcriptions, for comments as to whether the modern version is reconstructed or handed down, and for speculation as to whether one is more likely today to hear Hujia Shibapai or Da Hujia.

6. Luqi Xinsheng has no information about where it was published.

7. Attribution of this Hujia to Xu Shiqi
At the beginning of the melody (see VII/31) there is the statement "written by Xu Shiqi", but this is also the first page of Folio 3 of Luqi Xinsheng, and so the attribution should be to the book, not to the melody. On the other hand, according to QSCB, Chapter 7a4, Xu Shiqi was noted for creating qin melodies with lyrics.

8. Source of the Hujia Shibapai lyrics attributed to Cai Wenji (蔡琰 Cai Yan)
According to Xu Jian, QSCB, Chapter 6b1-2, the first publication of this version of the poem is in the 楚辭後語 Afterword to the Songs of Chu, compiled by Zhu Xi (1130-1200). However, elsewhere it is said that the earliest known publication was in the 12th century Yuefu Shiji, Folio 59, #3 (pp. 860-865), where it is said to be the original, and that it was later imitated by Liu Shang (late 8th c.). However, the Liu Shang poem, also a first person narrative, is known to have had some popularity during the late Tang, so it could well be that the poem attributed to Cai Wenji herself was the imitation. See Idema and Grant, p. 121ff. It is translated there as well as in Chang and Saussy, pp.22-30.

9. Tracing the 1597 version of Hujia Shibapai
A chart under Da Hujia traces Hujia Shibapai together with Da Hujia. As mentioned there, it seems likely that the 1597 version survives only in 1597 and 1611. This conclusion is based on analyzing the versions that have Cai Wenji's lyrics, as follows.

Zha's guide has Cai Wenji's lyrics in the following handbooks:

  1. Luqi Xinsheng (1597; VII/31)
  2. Qin Shi (1611; VIII/44; identical)
  3. Lixing Yuanya (1618; VIII/325; different melody, 9-string qin)
  4. Wuzhizhai Qinpu (1722; XIV/557; the lyrics come after the tablature: different melody that doesn't fit by the traditional method)
  5. Qinxue Renduan (1828; lyrics are apparently again separated from melody)
  6. Kumuchan Qinpu (1893; facsimile IV/10; like 1722)
  7. Qinxue Congshu (1910; not in QF; fac.9: tries to pair lyrics with 1722)

I have not been able to examine all the 19th century handbooks, but based on the above it seems likely that all the melodies there are related to 1722.

10. Published qin melodies with Hu Jia in their title
Zha Guide includes indices only for Da Hujia and Xiao Hujia. Three other surviving titles are said to be variations of Da Hujia: Shiba Pai, Hu Jia and Hujia Shibapai. Although these are all said to be variants of Da Hujia, as mentioned above the Hujia Shibapai played today, though somewhat related, is actually very different from Da Hujia; as for the version sung to Cai Wenji's lyrics (the focus of the present page), it is a completely different melody.

11. The Da Hujia set to lyrics in <1491 uses the same melody as in 1425, and the lyrics seem to be new and not naturally adaptable for singing.

12. Other qin songs on the Hujia story?
The Qin Shi Xu biography of Chen Shi says "今琴曲有胡笳十八拍詩所造也 present-day qin melodies include a Hujia Shibapai poem he wrote." There is some further information in Xu Jian's QSCB, Chapter 7, Qin Song Composers.

13. Traditional operas on the Hujia story
Traditional Chinese concern this story (see LXS) include Wenji Goes to the Desert (文姬入塞 Wenji Ru Sai) and Wenji Returns to Han (文姬歸漢 Wenji Gui Han). A modern opera by Bun-Ching Lam (Wenji: Eighteen Songs of a Nomad Flute, 2002) is mentioned elsewhere.

14. Reconstructions by Zha Fuxi and Chen Changlin
Zha Fuxi's transcription can be seen in Jinyu Qin Ge; Chen's was apparently not published until 2013 (further below). I have not examined them carefully enough to know to what extent they may have been influenced by each other..

15. Transcription of Hujia Shibapai by 王迪 Wang Di
As yet there do not seem to be any essays published comparing Wang Di's version with the reconstructions by Zha and Chen. The most easily available of her transcriptions seems to have been given general publication in 1982.

Whether from here or elsewhere, Wang Di's interpretation of this melody seems to have been quite well-known: in particular I have heard people sing the first section from memory, or recognise it when they hear it. However, when asked they have not been able to say where it came from ("traditional"?), and I have not yet found the earliest performance based on Wang Di's transcription, nor do I know when Wang Di first did her transcription. I have seen three publications that include it:

In all these transcriptions Wang Di has consistently changed non-pentatonic notes to make the music pentatonic. According to my study, most of these non-pentatonic notes are mi flats and do sharps. These occur with such consistency that they must have been intentional (see further comments, for example, under Shang mode and Wuyi mode).

The 1982 publication also has Wang Di's transcription of Sections 9 and 10 of the musically unrelated 1722 Hujia Shibapai; that 1722 volume appends Cai Wenji's lyrics after the tablature, so Wang had to pair them to the music according to her own method.

16. Opera version of the 1597 score
Perhaps 王迪 Wang Di herself (see above) was involved. Some incomplete instrumental versions can be heard online (see further details with commentary on the 1425 version), and occasionally Chinese friends have surprised me by singing the opening lyrics. However, they do not recall who arranged this version into an opera (or song cycle), or how complete it is.

17. Adaptations of the 1597 Hujia Shibapai for qin
In 2012 and 2013 I heard for the first time complete performances of the 1597 version for solo voice and qin. The two performances seemed somewhat different. Some details of this are as follows:

  1. On the double CD 沈德皓琴歌 Shen Dehao Sings Traditional Chinese Qin Songs (CD-2287S no date given) there are, in addition to a historical recording with orchestra of sections 1, 2 and 12, an undated complete recording of Shen Dehao singing with solo qin. It sounds as though the melody is sung pretty much as written in the qin handbook, but quite a bit of the fingering is changed (mostly through selective elimination), freeing the music of the restriction of one syllable per right hand stroke.
  2. In a live performance, on 20 December 2012 in Hong Kong, 王穎苑 Wang Yingyuan sang the entire lyrics in Suzhou singing dialect (though not in Suzhou singing style). She may have also sung the melody as written in the qin handbook, but she also seems to have eliminated some of the fingering, perhaps even more than in the recording. In any case, it was quite effective

I don't know whether Wang Yingyuan's version was based on Shen's interpretation, or whether they were simply both based on the original reconstruction by Zha Fuxi and/or Chen Changlin.

18. Transcription and recording by Chen Changlin
A preliminary investigation of his transcription suggests that Chen did not change as many of the non-pentatonic notes (most commonly 1♯, 4♯ and 3♭, all of which were typical of Ming dynasty idiom) as did Wang Di. However, in his recording (CD 4, track 9) many of these notes seem to be either changed or unclear.

19. Transcription by John Thompson (see first page)
My transcription, written in Encore, is 22 pages of staff notation (four lines to a page), with the original lyrics in Chinese characters, romanization (pinyin) and literal translation to English paired underneath the notes. In addition I hand wrote the original tablature underneath. My idea of having it performed would be to find a singer who would first try singing it according to my transcription, but then (having an understanding of this structure) develop her or his own interpretation. (There are currently a few copy errors in my transcription, e.g., in the vocal line of m.173.)

Regarding the relationship between the song line and the qin line, as described elsewhere, there seems to be no comment about this anywhere in old qin sources; my own feeling is that the result should be heterophonic: it is not so interesting if the voice simply sings over the qin line, thereby hiding the complex but delicate nature of the silk string qin sound (with nylon/metal or composite strings this is not such an issue, since their sound is much simpler). An additional important performance issue for the singer on this particular melody is the last note of each section. Sections #10-11 and #14-16 all end on 6 (la) and Section 17 ends on 1 (do). These are clear enough, but Section 18 ends on 6 over 3 (mi): which note should the singer select? And all the other sections (#1-9 and 12-13) end on 1 followed by an upward slide to 5 (sol) then a left hand pluck giving a low 3 (mi): how should the voice interpret this? (My current idea, not shown in the transcription, is for the singer either to sing only the 1, then take a breath as the qin slides then plucks; or to sing and hold 1, then slide up to 3, followed by the pause.)

With the whole piece being so word intensive from beginning to end, it might also be appropriate to add short instrumental sections, for example, playing the first phrase of each section solo before having the voice come in.

20. Hujia Shibapai as attributed to Cai Wenji: original lyrics
There are translations in Idema and Grant, p. 121ff; and in Chang and Saussy, pp.22-30. The original lyrics are as follows (compare the lyrics by Liu Shang):


  1. 第一拍

  2. 第二拍


  3. 第三拍

  4. 第四拍

  5. 第五拍
          (The qin version then repeats from "五拍....", with different music)

  6. 第六拍

  7. 第七拍

  8. 第八拍

  9. 第九拍

  10. 第十拍

  11. 第十一拍

  12. 第十二拍

  13. 第十三拍

  14. 第十四拍

  15. 第十五拍

  16. 第十六拍

  17. 第十七拍

  18. 第十八拍

Return to the annotated handbook list or to the Guqin ToC.