Spreading the news: The circulation of Periodicals in Crow’s Newspaper Directory of China

This essay is the continuation of our series devoted to Crow’s Newspaper Directory of China. In the previous essay, we analyzed the layout of periodicals. In this new instalment, we examined their circulation.

Note: An enriched version is available as a R Markdown document (enriched with the data and lines of code for replicating the visualizations).


In the preface of the directories, Crow made a tentative assessment of the total number of newspapers and newspaper readers in China. He estimated that the total number of copies of papers printed each day was less than 2,000,000 copies in 1933, but more than 3,000,000 in 1935 and 1937 (1,000 million a year)1. On this basis, Crow further estimated that there were between 9,000,000 and 18,000,000 newspaper readers in China (These statistics did not include non-daily periodicals and foreign-language newspapers).

It is not clear how he arrived at these figures. As he explained in the preface:

We have made a careful survey of the publication and distribution of Chinese language dailies and have arrived at the figures showing the number of newspapers circulated in each province (and Hongkong) as well as the number of newspapers circulated for each 10,000 population2.

However vague this statement, there is little doubt that Crow relied on the data he collected in the course of his work as an advertising agent and manager of Carl Crow, Inc. – the advertising agency he founded in 1918:

In compiling this table we have given careful consideration to the geographical distribution of the larger newspapers as shown by checked advertising returns we have kept over a period of years and covering many thousand keyed returns3.

Although most figures came from the publishers themselves, they were supposed to be fairly accurate. As Crow explained:

Most of the circulation figures published in this directory are supplied by publishers and although there are some exaggerations we believe the figures in the main are fairly accurate4.

It remained difficult to assess the actual number of newspapers readers. Few readers subscribed to more than one periodical, but each copy was passed on to several other readers (from five to ten):

The number of newspaper readers is of course greatly in excess of the above figures. There is little duplication of circulation in China, that is, it is rare for the individual to subscribe for more than one paper, and there are probably more readers for each copy printed than in any other country. Many newspapers after being read by the original subscriber are passed on to others and it is a common experience for replies to be received to advertisers months after the advertising has appeared5.

Let’s have a closer look at these statistics.

General statistics

In the preface of the 1933 and 1935 editions of the directory, Crow provided general statistics regarding the daily circulation of newspapers in each province and in proportion to the population of the province. The former most likely resulted from the aggregation of the individual circulation of all newspapers listed in each province, and the latter was probably obtained by computing the ratio between the total number of newspapers circulated in each province and the population of that province.

We compiled these statistics in a table that indicates for each province:

  • its standard name (contemporary pinyin) (Province)
  • the edition of the directory (Year)
  • the number of newspapers circulated daily (Newspapers circulated daily)
  • the population of the province (Population)
  • the number of newspapers circulated per 10,000 inhabitants (Newsp. per 10,000 population).

There were 22 provinces documented in 1933 and 23 in 1935. The list also includes three non provinces – Dalian, Hongkong, Macao – which we will have to set apart when comparing statistics across provinces.

Based on this data, we produced the two sets of maps and barplots.

Daily circulation by province (1933-5)

The barplots reveal huge discrepancies between provinces:

In order to better visualize spatial discrepancies over time, we relied on Geographical Information System (QGIS) to map the daily circulation of newspapers by province in the two directories. Based on the “Natural breaks (Jenks)” classification, we defined six thresholds:

As suggested earlier, coastal provinces, especially Jiangsu and Hebei and Guangdong, displayed the highest circulation figures. The maps show an almost linear gradient from the coast towards the interior, yet with a few exceptions. The circulation of newspapers was particularly low in the coastal province of Guangxi, whereas the central provinces of Sichuan and Hubei showed remarkably high circulation figures. Moreover, the circulation of newspapers in interior provinces progressed significantly between the two editions.

Circulation per population (1933-5)

If we consider the newspaper per population ratio, we observe a much more dramatic gap. People in the various provinces clearly had unequal access to the press. Readers in Hongkong and Dalian were the best supplied. Dalian ranked first in 1933 but Hongkong took the lead in 1935. This reversal, however, may reflect Crow’s difficulties in updating data in Japanese-occupied territories. Other provinces stood far behind the two top providers.

In 1933, the number of newspapers available for 10,000 inhabitants ranged from just one in rural provinces such as Gansu and Guizhou to 3000 in Dalian. The situation had barely improved by 1935. The mean and median had increased (39 and 461 respectively) and so had the minimum and maximum. There were 3 newspapers for 10,000 readers in Guizhou and 4 in Gansu, compared to 5393 in Hongkong.

How can we account for such discrepancies in terms of access between provinces? The scale of observation is a first factor we need to take into account. As cities, Dalian and Hongkong can not really be compared to province. Let’s remove them in order to get a clearer picture:

There are still huge discrepancies between provinces. Jiangsu ranked first with over 200 newspapers distributed per 10,000 inhabitants in 1933, and 300 in 1935.This may still appear a low figure compared to other countries. It implied that only a very limited fraction of the population could actually access newspapers. Yet as Crow explained in the preface, many newspapers, after being read by the original subscriber, were passed on to others so there were more readers for each copy in China than in any other country (Crow, 1935, p.3).

The next best equipped provinces after Jiangsu were Hebei (due to the presence of two major newspaper providers: Beijing and Tianjin), Macao (due to its smaller size), the Japanese occupied territory of Manchukuo, and a handful of southern, coastal provinces (Guangdong, Yunnan, Zhejiang). Yunnan made significant progress between the two editions of the directories.

In order to better visualize spatial discrepancies over time, we relied on Geographical Information System (QGIS) to map the the circulation ratio by province in the two directories. Based on the “Natural breaks (Jenks)” classification, we defined five thresholds:

As on the previous ones, these maps show a gradient from the coast to the interior. The ratio, however, remained relatively low in most central provinces. Only Shanxi, Shaanxi and Jiangxi showed some progress between the two editions. It appears that the increase in daily circulation was not sufficient to meet the needs of the reading population. It was perhaps more common to recycle newspapers in the interior than it was on the coast.

Alternative sources

In the 1937 directory, Crow relied on additional statistics to demonstrate the general growth of the daily press and newspaper readers in China in the early 1930s: importation of newsprint (i.e. cheap paper used to print newspapers) and the number of periodicals transmitted by the Chinese Post office in 1937:

Importations of newsprint in China (1931-35):

1931 51327
1932 55700
1933 65266
1934 77510
1935 144160
Although some newsprint may be used in the printing of cheap books, almost all of it was consumed by the daily newspapers, so that newsprint imports indeed gave a fairly accurate indication of the growth of the daily press in China.
Newspapers transmitted through the Chinese Post Office (1922-34):
1922 80000000
1931 163082300
1932 186478300
1933 180659300
1934 169667800

Although the post office statistics also included weekly and monthly periodicals, daily newspapers represented 90% of the total. Transmission of newspapers through the Post Office, however, accounted for only a part of the total circulation of newspapers. Local subscribers received their papers through delivery by coolie. Crow estimated that 50% of large newspapers in Shanghai were delivered to outport subscribers through the Post office, and the remaining 50% to local subscribers by coolie service. Moreover, the smaller papers in Shanghai, as well as many of the provincial dailies published in 120 cities, were restricted to a circulation which was almost entirely local, and therefore not included in the Post Office statistics (e.g. Shandong). Note: The slight drop in the figures for 1932-33 was due to the discontinuance of the Chinese Postal Service in Manchuria (Crow, 1937, p.v-vii).


For this research, we rely on the following information from the directory:

Variables related to circulation:

  • Circulation_range: a binary variable that indicates whether the circulation varied or not (fixed/varying) (P)
  • Audited: a binary variable that indicates whether the circulation figures resulted from an external audit or came from the editors themselves (Audited/Publisher’s statement);
  • Circulation_avrg: average number of copies circulated (R);
  • Circulation_min: minimum number of copies circulated (R);
  • Circulation_max: maximum number of copies circulated (R);
  • Circulation_threshold (P): an additional categorical variable derived from the average circulation, which defines eight thresholds in order to make sense of the wide range of circulation figures.

Other variables used for interpretation:

  • Year: Year when the directory was published
  • Periodicity: the frequency of publication. In order to facilitate further analyses, we lumped together rare categories of periodicals, resulting in five main groups (Periodicity_fct): Daily, Tabloid, Weekly, Monthly, Intermediate (bi/semi weekly and monthly = 37), Quarterly and Annuals (AQ = 24).
  • Language: the language of publication, which we also re-classified into four main groups (Language_fct): Chinese, English, Japanese, other;
  • City: city in which the periodical was published;
  • Population: population of the city (as estimated by Crow).

Most information consists in “raw” data drawn directly from the directories (R), but some have been produced to enable historical interpretation (P).

Let’s start with a general overview of the average circulation of periodicals, all categories and languages included.



Circulation figures are unknown for 56 periodicals in 1931 (15%), and 24 in 1935 (3.4%), which suggests greater transparency from the editors and better access to information in the press industry. As Crow explained in the preface, most figures, however, came from the publishers themselves (“Publishers’ statement”). Only twelve periodicals were certified by an external audit firm (“Audited”) (8 in 1931, 9 in 1935).

Except for the Chinese magazine Liangyou (Young Companion), audited figures referred exclusively to English-language periodicals. Except for the Hongkong-based South China Morning Post and Hongkong Telegraph, all were published in Shanghai. As China’s business and financial center, Shanghai hosted many accounting firms. Since it was also the national center for the publishing and press industry, there was a stronger pressure on newspaper editors in Shanghai to provide accurate data than there was in any other city in China.


As the histograms reveal, the average circulation of periodicals increasingly approximated a normal distribution:

The number of copies ranged from 220 to 150000 in 1931, from 220 to 150000 in 1935, with a slightly decreasing mean (7310 to 6447) and median (3000 to 2950) between the two directories.


Based on the histograms, we defined seven thresholds to facilitate the interpretation of circulation figures:

The periodicals whose circulation ranged from 1,000 to 5,000 represented the majority in the two directories (52%). Those with extreme circulation figures remained a minority. The smallest ones represented less than 10% and declined between the two editions (from 8% to 6%). They included a variety of publications in terms of periodicity and language, from Chinese dailies, tabloids and semi-weeklies, to English monthlies and quarterlies.

The most widely circulated periodicals amounted for less than 5% but progressed during the same period (from 3 to 13). Only three newspapers distributed 100,000 copies or more. The two competing Shanghai giants Shenbao 申報 and Xinwenbao 新聞報 held the record (150,000), while the Shishi xinbao 時事新報 claimed a daily circulation of 100,000 copies.

However huge these figures may appear in comparison to other periodicals in China, they remained much lower than those enjoyed by the largest newspapers in Western countries. The Times in Great Britain, for instance, reached almost 3,000,000 copies during the same period.

The categories in between represented less than 10% each. Smaller publications circulating 500 to 1,000 copies experienced the most remarkable growth (from 40 to 105, 13% to 15%). Higher figures (5,000 to 50,000) progressed at a slower pace.

Multivariate effect

How did circulation relate to the the frequency, the language and the place of publication?

As the plots reveal (see below), there were huge variations between periodicals across and within categories (periodicity), languages and cities. We assume that circulation primarily depended on the frequency and language of publication, but other factors may also come into play.


How far did the number of copies depend on the frequency of publication?

There is no clear pattern from the histograms. As the boxplots reveal, each category, especially the most statistically important (dailies, weeklies, monthlies) presented a wide spectrum of possible circulation:





Clearly, we need to examine each category (periodicity) separately and take into account multiple factors especially the language, the place of publication (the city and its population) and the relative longevity of periodicals (age/year of establishment):

  • Language: Since Chinese readers far outnumbered English and other foreign readers, we may reasonably assume that Chinese publications in a given category (periodicity) enjoyed larger circulation than their foreign counterparts.
  • City and population: we assume that daily newspapers in large metropolitan centers, especially on the eastern coast, enjoyed a wider circulation than newspapers published in smaller towns in the interior.
  • Age (year of establishment): we hypothesize that older publications relied on a stronger reputation, and therefore enjoyed a wider circulation, than younger competitors in the same category.

For each category of periodical, we will proceed in several steps:

  1. First, examine the general distribution based on the average circulation (histogram) and circulation threshold (barplot).
  2. Second, we will try to assess the influence of age (scatter plot).
  3. For dailies specifically, we will also analyze the distribution by language (boxplot) and city (boxplot) in order to examine whether there was a relation between circulation and population figures (scatter plot).
  4. For weeklies and monthlies specifically, we will compare the difference in circulation between English and Chinese periodicals.

Let’s start with the most popular category: daily newspapers.



As the histograms reveal, the circulation of daily newspapers followed an almost normal distribution:


In 1931, the circulation of daily newspapers ranged from 220 – Chefoo Daily News – to 150,000 – Shenbao and Xinwenbao. Information is missing for 47 newspapers (19%). Those circulating from 1,000 to 5,000 copies represented the majority (107 titles, 55%), among which 53 (28%) distributing 1,000 to 2,500 copies and 50 (26%) from 2,500 to 5,000. Larger papers (5,000 to 10,000) amounted for 18% (34). The smallest (under 1,000) and largest figures (over 10,000) represented a minority (15% and 14%, respectively). The Chefoo Daily News was the only paper to circulate less than 500 copies per day.

The situation had not changed much by 1935. Newspapers with circulation from 1,000 to 5,000 still represented the majority (165, 59%). The lower slice (1,000-2,500) had increased to 32% (140 titles), whereas the higher (2,500-5,000) remained stable (119, 27%). The most significant change was the growth of extreme figures. Small papers under 500 copies now amounted for 4% (16 titles) and we now had nine “giants” circulating over 50,000 copies per day (2% only). In between, only 22% distributed from 5,000 to 50,000 copies, compared to 30% in 1931. Smaller papers (500 to 1,000) had grown in number (26 to 63) but remained stable in proportion (13-14%). Information is missing for just 14% (3%), which confirms the greater transparency from newspaper editors.


Was there a relation between the longevity and the circulation of newspapers? We hypothesize that older newspapers had more time to build their reputation, and therefore enjoyed a larger circulation. In a nutshell, the older the periodical, the stronger its reputation and the larger its readership. Did the directory support this assumption?

From the scatterplots, there was no clear relationship between age and circulation. Although a few, almost centennial newspapers displayed high circulation figures (dots on the top right hand corner), we also find long-established ones with much lower coverage (bottom right). Conversely, although the youngest, most recently established papers (less than a year) usually showed lower circulation figures (less than 10,000), many dailies established ten or five years earlier were already widely distributed (more than 10,000 and even close to 100,000 copies).

A linear regression analysis (lm), setting the average circulation as the outcome variable (y) and age as the explanatory variable (x), however, reveals that age had a stronger influence on the circulation of newspapers than the plot suggests (p value: < 1.526e-08 in 1931, 5.311e-11 in 1935). The number of copies they distributed did increase as newspapers grew older. More precisely, each time they gained one year, their circulation increased by 626 copies in 1931 and 454 in 1935.

The circulation of the youngest periodicals may vary from 700 (Wise Press) to 6,000 copies (New Hankow Daily Press). The latter was able to build a strong readership in a limited period of time. On the opposite, the two British pioneers – North-China Daily News (NCDN) and Hong Kong Daily Press (HKDP) maintained relatively low circulation figures (less than 10,000, under 5,000, respectively).

We notice, however, that their circulation increased between the two editions. The HKDP had gained 1,000 readers in four years and the NCDN almost twice this number. Moreover, the language itself did not account for disparities in circulation. Among the three oldest newspapers, the only Chinese-language Chinese Mail 華字日報 (Huazi ribao) – based in Hongkong Chinese Mail – stood between its two English predecessors (5,000) and far below its Chinese junior Shenbao (30 more copies circulated daily). The difference in size between the two cities (Hongkong had a population of just 513,000 compared to three millions in Shanghai) may account for such a gap.

Neither age and place alone, however, did account for variation in circulation. Many factors intermingled in complex ways to account for the varying popularity of newspapers. Let’s focus on the most obvious one – language.


Not surprisingly, Chinese dailies displayed the largest spectrum. The highest median, however, went to Japanese newspapers (7,000 compared to 3,000 for Chinese in 1931, 2,600 in 1935). The median circulation of Russian and English dailies was very close to the Chinese. These figures reflect differences in the structure of the daily press across languages.

As we demonstrated in a previous essay, the Chinese press presented a wide spectrum of newspapers with a very skewed distribution (from “mosquito” papers to giants). Chinese dailies could be found all over the country. Every town of enough importance had its own paper. By contrast, the Japanese and the Russian presses consisted in a limited number of large newspapers catering to their respective nationals concentrated in a few treaty ports or colonies (Dairen, Harbin).

English newspapers presented another specific case. The English press addressed not only British or American readers, but foreigners from various countries spread all over China. They also increasingly appealed to an elite of Chinese readers who had received a foreign education abroad or in a few selected missionary schools in China – Anglo-Chinese College in Fuzhou and Guangzhou, St John’s University in Shanghai, Qinghua in Beijing, to name but a few6.

The French and German press showed the lowest circulation, which reflected the comparatively small size of the French and German communities in China.


Did the circulation of newspapers depend on the nature and importance of the city in which they were published?

Large metropolitan centers with a multinational population topped the list, especially the three major treaty ports (Shanghai, Tianjin, Guangzhou) and colonies (Dalian, Shenyang, Hongkong). It was in these cities that we found the highest circulation figures and the largest scope. The former and current national capital – Beijing and Nanjing – came next. Provincial capitals in interior provinces, such as Kunming, superseded important urban centers on the coast, such as Hanghzou. Newspapers in smaller cities and towns enjoyed lower and usually more concentrated circulation figures, with a few exceptions (Zhenjiang, Wenzhou), which presented a lower but wider spectrum.


From the scatterplots below, there was no clear relation between population and circulation:

Whatever the size of the city, the dots representing the circulation are stretched along the y axis. Circulation figures may increase by a factor of 100 in cities with a population of 100,000 or 500,000, for instance. We observe a patter of higher concentration in smaller and larger cities alike (100,000 and 1,000,000) but larger population alone did not account for higher circulation figures.



There were only two tabloids for which we know the circulation in 1931 – the Chungkung Evening Post (重慶晚報 Chongqing wanbao) and the Local Noon Voice (市聲午報 Fusheng wubao). Both were published in Chongqing, Sichuan province. Although they were of recent creation (both were established in the early years of the Nanjing regime, 1928 and 1931, respectively), they already distributed more than 3,000 copies daily (4,000 and 3,214 respectively).

Due to the scarcity of information in the first directory, we will focus on the 1935 edition. As shown on the histogram, the distribution was very skewed. The number of tabloids decreased as the circulation increased. Small tabloids with low circulation dominated. There were very few tabloids with circulation close to or over 30,000, and none in between (15,000 and 30,000).


Tabloids circulated from 400 to 35,700 copies, with a mean of 6,900 and median 3,000. Small tabloids circulating 500-1,000 copies represented the largest share (23%), followed by intermediate (2,500-5,000) and larger ones (10,000-50,000) (21% each). Tabloids that circulated 5,000 to 10,000 copies represented 13%. The two remaining categories (less than 500 and 1,000-2,5000) amounted for 11% each.

The highest figures (over 25,000) refered to major urban centers, especially the former imperial capital Beijing and the new capital of the Nationalist regime Nanjing. The most widely circulated was the Min Sun Pao (Press of the People) (民生報 Minshengbao) established in Nanjing in 1927 (35,700 copies). Next we found two Shanghai tabloids – The Holmes (福爾摩斯 Fuermosi) (established 1924) and The Crystal (晶報 Jingbao) (1919) – which distributed 32,000 and 30,000 copies, respectively. Finally, two Beijing-based tabloids – Shao Pao (The Truth) (實報 Shibao) (1928) and Chun Chiang Pao (Strong Group Press) (羣強報 Qunqiangbao) (1921) – circulated 28,000 and 27,000 copies, respectively.

The three smallest tabloids (400 copies) were all based in Zhenjiang (Jiangsu) and had been established since 1930. Tabloids with circulation from 500 (excluded) to 1,000 (included) comprised a variety of publications.


From the scatterplot, there was no clear relation between age and circulation.

A linear regression analysis (with circulation as the outcome variable and age as the predictory variable), however, reveals that age had a stronger impact on the circulation of tabloids than the plot suggests (p-value: 1.441e-05). Each time they grew one year older, their circulation increased by 1486 copies.

Three among the oldest tabloids (over 10 year-old) – The Holmes (11 y.o.), The Crystal (16 y.o.) and Strong Group Press (14 y.o.) – enjoyed the highest circulation (30,000 or more), whereas the three least circulated (400) were also among the youngest (established between 1930 and 1934).

Among the youngest and least circulated, we found pioneers such as Petty News (Shao Jih Pao) (小日報 Xiao ribao), established in 1919, which distributed less than 3,000 copies 16 years later. On the opposite, more recently-established titles – such as Mei Jih Sin Wen (新閒 Xin Wen) or New Shanghai (新上海 Xin Shanghai) – both established in 1934, were able to distribute more than 10,000 copies during the first year of their existence.


How did the place of publication impact on the circulation of tabloids?

Beijing tabloids enjoyed the highest median circulation (27,000), far ahead other cities. Shanghai came second (5,000), followed closely by Chongqing and Wuxi. Nanjing, Hongkong and Zhenjiang came last (500). Nanjing displayed the widest spectrum – from 500 copies for the Morning News (朝報 Chenbao) to 35,700 for the Press of the People (民生報 Minsheng bao), before Shanghai (from 1,000 to 32,000) and Beijing – from 2,000 (Mining News 礦業新閒) to to 28,000 (The Truth 實報).


As the scatterplot suggests, the circulation of tabloids did not depend on the size of the city:



In contrast to the normal distribution of daily circulation, weeklies presented a different picture. We found a larger number of small weeklies (<1000 copies) and few with large circulation figures (over 10,000):

In a word, the larger the circulation, the lower the number of weeklies. There were notable exceptions, however, and significant changes between the two directories.


In 1931, weekly circulation ranged from 500 to 150,000 (Xinwenbao Pictorial). Information is missing for three periodicals. Lowest figures (under 500 copies) represented the largest share (15 titles, 28%). Medium circulation (2,000-5,000 and 1,000 to 2,500) amounted for 23% (12 titles) and 17% (9) respectively. Only 3 weeklies (6%) were able to distribute more than 10,000 copies. The remaining categories (500-1,000 and 5,000 to 10,000) represented 13% (7 titles) each.

The situation had changed significantly by 1935. Weekly circulation now approximated the normal distribution we observed for dailies. Since the Xinwenbao Pictorial was no longer included in the list of weeklies, the maximum had lowered to 65,000 – Radio Movie Weekly 電聲電影週刊 (Diansheng dianying zhoukan). Intermediate circulation (500-1,000, 1,000-2,500 and 5,000-10,000) had grown to 20% each. The smallest circulation (under 500) had dropped to 10%, and the 2,500-5,000 category to 16%. The largest circulation (over 10,000) now represented 15% (11 titles). Information is missing for 5 periodicals (7%).


The scatterplots show no clear relation between longevity and circulation, though younger periodicals tended to enjoyed larger circulation figures:

As we demonstrated in a previous essay, there were two main generations of weeklies. The first generation comprised English-language weeklies established in the late 19th century, such as the pioneering North-China Herald, whereas the second generation referred to popular Chinese magazines that flourished in the early 1930s, such as the New Life Weekly (新生活週報 Xinshenghuo zhoubao). Due to language constraints, the former enjoyed lower circulation than the latter. Established in 1934, the New Life Weekly was able to distribute 60,000 copies the year after, whereas the North-China Herald maintained a circulation of 2,000 copies per week (7,950 for its Sunday edition).


As we observed in a previous essay, there were only Chinese- and English-language weeklies in 1930s China

From the boxplots we can make two major observations. Chinese weeklies displayed a much wider spectrum but the two languages approximated the same median (2,000 in 1931), though the gap widened in 1935 (2,650 for English vs. 3,000 for Chinese).


Did the place of publication impact of weekly circulation?

After the disappearance of the Mukden Pictorial published in Shenyang in 1931, Shanghai invariably ranked first by the range and importance of its weekly circulation figures. Second-tier cities and towns that did not appear in the list of dailies came next (Taiyuan, Datong and Taoyuan). Larger urban centers presented a wider spectrum, reflecting the size of their respective market.


How did circulation relate the population of the city?

The scatterplots deny any relation between circulation and population. The size of the city did not clearly affect the circulation of weeklies.



The circulation of monthlies approximated a normal distribution in the two directories:


The figures ranged from 300 to 45,000 in 1931 and 50,000 in 1935. Information is missing for just one periodical in 1931 (2%) and 5 in 1935 (6%). Although the median remained the same, the distribution changed significantly between 1931 and 1935. The most important change was the remarkable growth of medium circulation (1,000 to 2,500 copies). Their share increased from 20% to 29% (9 to 22 titles). Larger monthlies (5,000 to 10,000 copies) rose from 11% to 16% (5 to 12 titles). Lower circulation (500-1,000) increased slightly from 13% to 16%. Other categories declined in proportion, especially the lowest figures (under 500) that fell from 9% to 4%. Intermediate circulation (2,500-5,000) decreased from 31% to 24%. The most widely circulated monthlies (over 10,000) declined from 18% to 12%.


Did the longevity of monthlies impact on their circulation?

There was no clear relation between age and circulation. We notice that the most widely circulated monthly in 1931 – The Eastern Miscellany (東方雜誌 Dongfang zazhi) – was also one of the oldest (established in 1903). Conversely, the youngest monthlies displayed the lowest circulation. But we also found long-established journals such as the China Medical Journal (1886) with very low circulation figures (2000), while younger magazines (under 10 y.o.) rapidly gained in popularity. The Young Companion (良友 Liangyou) established in 1925, for instance, was able to distribute 40,000 copies five years later.

The relation had reversed by 1935. Some of the most recently established monthlies were able to circulate over 30,000 copies. Two monthlies established since 1930 – Juvenile Student (中學生 Zhong xuesheng) in 1930 and Shun Pao Monthly (申報月刊 Shenbao yuekan) in 1932 – already distributed 50,000 copies.

By constrast, the oldest periodicals (over 30 years old) were among the least circulated (less than 3,000 copies per issue). Among them, we mostly found professional and intellectual journals such as The China Medical Journal (est. 1886) and The True Light Review (真光雜誌 Zhenguang zazhi) (est. 1902), which addressed a highly educated and specialized readership (2,000 and 1,350 copies, respectively).

In sum, the gap between circulation figures reflected the segmented nature of the monthly market (specialized, medical journals vs. popular, general magazines). The content and readership of monthlies (specialized or popular) was the main factor that accounted for the observed disparities.


Not surprisingly, Chinese monthlies displayed a wider range (from 500 to 50,000, median: 3,000) and higher figures than their English counterparts than their English counterparts (from 300 to 3,000, median: 1,500).


How did the circulation of monthlies relate to their place of publication?

The Shanghai monthly press invariably presented the widest range and the highest circulation figures. The national capital Nanjing came second but its median decreased between the two directories and the spectrum widened between the two directories, ranging from 1,000 for the Communication Magazine (交通雜誌 Jiaotong zazhi) to 10,000 for the Current Events (時事月報 Shishi yuebao) in 1935.

The former capital Beijing produced only two medical journals in 1935. Their circulation varied from 500 for The Living Medical Journal 唯生醫學 (Weisheng yixue) (est.1932) to 3,000 copies for the Chinese Medical Journal (est. 1914).

n Hongkong, the official organ of the local Chinese General Chamber of Commerce (香港華商總會月刊 Xianggang huashang zonghui yuekan) enjoyed the largest circulation (2,000 copies). The lowest figures referred to the academic journal St. John’s Review (300 copies).

The remaining cities published fewer monthlies but some of them acquired a relatively high popularity, such as Happy Garden (樂園 Leyuan) (3,000 copies) in Chengdu or the Kwang Chi Medical Journal (廣濟醫刊 Guangji yikan) in Hanghzou (2,500).


The scatterplots reveal no clear relation with the population of the city. Monthly circulation depended less on the size of the city (measured by its population) and more on the importance of their intended readership. Highly specialized journals enjoyed lower circulation than popular magazines, even if the former were published in larger cities than the latter.

Intermediate Periodicals (bi/semi)


As evidenced by the histograms, intermediate periodicals did not show any clear circulation pattern:

The detailed histograms reveal that semi-weeklies generally enjoyed a lower circulation than semi-monthlies:

Semi-weeklies presented the widest scope (from 300 to 30,000 copies), which reflects their greater statistical importance and the geographical dispersion of their place of publication. They were published not just in large urban coastal centers, such as Guangzhou, but also in smaller towns in the interior, such as Zixing (Hunan) or Chongde (Heilongjiang).

Semi-monthlies presented higher figures (none of them distributed less than 1,000 copies) but this mostly reflected their concentration in Shanghai and the popular nature of their contents (children, women and students’ magazines.

The only bimonthly listed in the directory – The Journal of Electrical Engineering (電工 Diangong), based in Hangzhou – distributed no more than 1,000 copies.


The detailed histograms confirms our previous observations:

Semi-weeklies were represented in all six categories, with a decreasing presence as we move towards higher figures. Semi-monthlies enjoyed the highest circulation, whereas bi-weeklies confined to medium figures (1,000 to 5,000). Let’s examine each category separately.

The first edition of the directory listed four biweeklies, all established one year earlier (1930). Their circulation ranged from 1,100 (Liaoning Educational Journal) to 4,000 (蔬肝報). Two were based in Hongkong – Soo Kun Pao (蔬肝報 Shugan bao) and Sweet Country (Wun Yun Haung) (温柔鄕 Wenzhouxiang), one in Changzhou (Jiangsu) – Steel Press (鋼報 Gangbao) and one in Shenyang –Liaoning Educational Journal (遼甯敎育公報 Liaoning jiaoyu gongbao).

The spectrum had dramatically widened by 1935. We now found 33 periodicals that circulated from 300 to 45,000. The mean was 8,959 and the median 2,150. They were quite evenly distributed across categories (from 12% to 19%). The lowest (under 1,000) and highest figures (over 10,000) took the lion’s share (38% and 19%, respectively). Periodicals in between (1,000 to 5,000) totalled 28%.


As the scatterplot reveal, there was no clear relation between longevity and circulation:

The oldest semi-monthly – Children’s World (established in 1923) – enjoyed the highest circulation (36,000) but the two 14-y.o. semi-weeklies – Tsungming Pao 崇明報 and Ching Kiang Chuen – distributed only 2,000 and 1,000 copies, respectively.

Conversely, the two-year-old semi-weekly Tzeshing People’s Press (資興民刊 Yilin xinzhi) distributed only 300 copies, whereas the New China Fortnightly (新中華 Xin Zhonghua) and Thousand Years Semi-Monthly (千秋半月刊 Qianqiu banyuekan) – two semi-monthlies established the same year – were able to circulate from 10,000 to 30,000 copies.

In sum, age alone did not account for the popularity of intermediate periodicals. We should pay attention to the exact frequency of publication and the size of the city where they were published. Less frequent periodicals (semi-monthlies) in large cities (Shanghai) generally enjoyed higher circulation than more frequent periodicals (semi-weeklies) published in smaller towns (Zixing).


As we observed in a previous essay, intermediate periodicals appeared exclusively in Chinese, except for the English-language biweekly China Truth. Based in Guangzhou, it distributed 2,300 copies – a higher figure than the median circulation of Chinese periodicals in the same category:


How did the circulation of intermediate periodicals relate to the city of publication?

With only two biweeklies in 1931, Hongkong ranked first by its circulation scope and median. In 1935, Chongqing showed a higher median than Shanghai but the latter displayed a wider spectrum (from 500 to 40,000). The reamining cities individuall presented fewer diversity but there were huge disparities between their respective circulation, with figures ranging from 300 in Zixing (Hunan) or Chongde (Heilongjiang), to 3,000 in Changzhou (Jiangsu).


As the scatterplots reveal, the circulation of intermediate periodicals did not directly depend on the size of the city (measured by its population):

Annuals & Quarterlies


Annuals and quarterlies presented a wide circulation spectrum, ranging from 400 to 8,000 copies:

The highest figures (more than 3,000) grew significantly between the directories. Annuals generally enjoyed a wider circulation than quarterlies (5,000 copies in 1931, no less than 3,000 in 1931). Since they implied a huge investment in time and money and heavy costs of production, they had to rely on a strong readership.

The detailed histograms confirm what we have just observed:


The threshold barplots present a clearer picture:

The detailed barplots refine this observation:

By 1935, the mean circulation had decreased while the median had increased, resulting in greater disparities between periodicals. Those which circulated between 2,500 and 5,000 copies experienced a significant growth (from 17% to 44%), at the expense of other categories, especially the lowest figures (under 500) which dropped from 33% to 11%. Figures between 1,000 to 2,500 dropped from 33% to 22%, and the largest (over 5,000) from 17% to 11%.


The scatterplots suggest some relation between age and circulation in 1931, but not in 1935. We can hardly generalize, however, given the paucity of information:

The oldest and most widely distributed annual was the Directory and Chronicle (5,000 copies), established in 1862. The youngest and least popular was the Hongkong Naturalist – an illustrated quarterly established in 1930 which catered to a niche readership of botanists in Hongkong and South China.

The 1935 directory presented a more complex picture, with no clear relation between circulation and longevity. The most popular quarterly – Tungchi Acta Medica (3,250 copies) – was also the youngest in the list (founded in 1931). By contrast, the oldest one – Nursing Journal Of China established in 1912 – did not exceed 1,500 copies, while the more recent Quarterly Journal of Economics of the Chinese Economic Society (經濟學季刊 Jingjixue yuekan) – established in 1930 – already distributed 2,000 copies.

The readership of the Hongkong Naturalist had barely grown since 1931 (from 500 to 600). Relying on a strong and growing community of Chinese engineers, the Journal of the Chinese Institute of Engineers (工程 Gongcheng) founded in 1922 was able to circulate 3,000 copies thirteen years later.


As we observed in a previous essay, annuals and quarterlies were published in Chinese and English only, including a few bilingual publications:

Chinese quarterlies enjoyed the highest median, which reflected two major professional journals – Tungchi Acta Medica and the Journal of the Chinese Institute of Engineers. English publications, however, displayed a larger spectrum, including such diverse periodicals as the reference Directory and Chronicle and the highly specialized Hongkong Naturalist. The bilingual Nursing Journal of China stood in between.


As we showed in a previous essay, the production of annuals and quarterlies was concentrated in four major cities:

As the main publishing center in China, Shanghai presented the highest median in the two directories and the widest spectrum in 1935 (from 2,000 to 8,000). In 1931, Hongkong and Bejing displayed the largest scope after Shanghai (500 to 5,000 and 400 to 2,500, respectively), while Nanjing ranked second by the median circulation (1,500) in 1935.


The scatterplots reveal no clear relation between the circulation of annuals or quarterlies and the population of the city where they were published:

These “unrooted” publications appealed to a large readership that extended far beyond their local site of production. Reference directories such as the Directory and Chronicle circulated more broadly in Asia and even Western countries.


The figures we have presented so far refer to the average circulation. For six periodicals, however, the directories gave an estimated range instead of an exact figure. Either the circulation may vary from one issue to the other, either the publishers were not able to assess the exact number of copies distributed. Among the six periodicals, we found three weeklies (two Chinese and one English), two monthlies (one Chinese and one English) and one Chinese quarterly (the medical journal published by the Medical college at Tongji University in Shanghai). Except for the Hongkong-based weekly The Critic, all were published in Shanghai.

The scatterplot gives a better sense of how far the circulation figures may vary:

The range of variation remained narrow. The English weekly Cathay Magazine (華懋雜誌 Huamao zazhi) presented the largest variability (ratio 1.5) with a circulation ranging from 1,000 to 1,500 copies. The Chinese weeklies – People’s Talk (Jen Yen Chow Kan) 人言週刊 (Renyan zhoukan) and English Weekly 英語週刊 (Yingyu zhoukan) – showed a ratio of 1.33 (from 30,000 to 40,000 copies). The Chinese monthly Child Education 兒童教育 (Ertong jiaoyu) presented the same ratio but half the number of copies (15,000 to 20,000). The variation was barely significant (1.167) for the The Critic, whose circulation varied between 600 and 700 copies per week.

Periodicity was the primary factor accounting for variation in circulation.

Concluding remarks

Due to low literacy rates, the fragmented nature of the market, and the practice of recycling newspapers among several readers, circulation figures remained relatively low in Republican China, compared to Europe or the United States at the same period.

As we showed in this essay, the circulation of periodicals in China resulted from the complex intermingling of multiple factors. Neither language nor periodicity alone could account for variation in circulation. The frequency of publication most strongly determined the range of copies, whereas language impacted primarily on their number.

Chinese dailies presented the broadest spectrum, from just 220 copies for the small Chefoo Daily News to 150,000 for two Shanghai giants Shenbao and Xinwenbao. The English press presented a wider distribution than other foreign languages, reflecting the multinational profile of its readership.

Longevity affected the circulation of dailies and tabloids only. As they older, daily newspapers and tabloids generally gained in circulation. By contrast, there was no direct relation between age and circulation for other categories of periodicals. The broad categories of weekly and monthly, in fact, referred to two very different kinds of publication. Whatever their age and reputation, the pioneering English-language periodicals established in the late 19th century could not compete with the Chinese popular magazines that popped up in the early 1930s. Neither could the highly specialized journals that addressed a niche readership of academic and professional readers. Intermediate periodicals were of too recent creation to make age a relevant factor in their circulation. Annuals and quarterlies were of too minor statistical importance to allow for any generalization.

The place of publication had no direct impact on the circulation of periodicals. Large urban centers on the coast certainly relied on a larger and more diverse market, based on the sheer size of their population, but they also offered a wider range of periodicals, which increased competition between publishers and market segmentation.

In a future essay, we propose to rely on multiple correspondence analysis (MCA) to bring together and correlate in a more systematic way the various factors that we have examined separately so far, in order to better illuminate their complex interplay.

Before we end this essay, one last intriguing thing to notice is how periodicals of comparable size often presented the exactly same figures, such as the two Shanghai giants Shenbao and Xinwenbao, which both claimed 150,000 copies daily. As for advertising rates, publishers were careful to aligned their statement in order not to “lose face” in face of their competitors7. They will be the focus of the next essay.


Bacon, C.A. “Advertising in China.” Chinese Economic Journal and Bulletin 5, no. 3 (September 1929): 754–67.
Carl Crow, Inc., ed. Newspaper Directory of China, Including Hongkong. Shanghai: Carl Crow Inc., 1933.
———. Newspaper Directory of China (Including Hong Kong). Shanghai: Carl Crow, Inc., 1935.
———. Newspaper Directory of China (Including Hongkong) & Advertising Manual: An Annual Data Book and Technical Record for All Newspapers and Periodicals in China. Shanghai: Carl Crow, Inc., 1937.
Crow, Carl. “Advertising and Merchandising.” In China. A Commercial and Industrial Handbook, edited by Julean Herbert Arnold, 191–200. Washington, Govt. Print. Off., 1926.
Sanger, J. Advertising Methods in Japan, China, and the Philippines. Washington: Govt. Print. Office, 1921.


  1. Crow, Newspaper Directory, “Preface”, 1933, 1935, p.2 ; 1937, p.vii []
  2. Crow, Newspaper Directory, “Preface”, 1933, 1935, p.2 []
  3. Crow, Newspaper Directory, “Preface”, 1933, 1935, p.2 []
  4. Crow, “Preface”, Newspaper Directory, 1933, p.2, 1935, p.5 []
  5. Crow, Newspaper Directory, “Preface”, 1933, 1935, p.3 []
  6. See “A circulation census. Increase in Chinese readers: Distribution Statistics,” North China Herald, May 24, 1933 []
  7. Bacon, 1929, p.755, Crow, 1926, p.196-7, Sanger, 1921, p.61-62 []

Cécile Armand

Ancienne élève de l'Ecole Normale Supérieure (ENS) de Lyon (2006), agrégée d'histoire (2009), docteure en histoire (2017), postdoctorante à Stanford University (2017-8) puis Aix-Marseille Université dans le cadre du projet ERC “Elites, Networks and Power in modern China” (2018-). Ma thèse soutenue à l’ENS Lyon en juin 2017 proposait une histoire spatiale de la publicité à Shanghai (1905-1949). Mes recherches actuelles portent sur l’émergence des experts, “l’influence” américaine, la naissance de la presse et d’une “opinion publique” en Chine moderne. Dans une démarche d’histoire sociale, mes centres d'intérêt englobent aussi l'histoire spatiale, les outils et méthodes numériques et l'épistémologie des sciences sociales en général.

More Posts - Website

Follow Me:

Cécile Armand

Ancienne élève de l'Ecole Normale Supérieure (ENS) de Lyon (2006), agrégée d'histoire (2009), docteure en histoire (2017), postdoctorante à Stanford University (2017-8) puis Aix-Marseille Université dans le cadre du projet ERC “Elites, Networks and Power in modern China” (2018-). Ma thèse soutenue à l’ENS Lyon en juin 2017 proposait une histoire spatiale de la publicité à Shanghai (1905-1949). Mes recherches actuelles portent sur l’émergence des experts, “l’influence” américaine, la naissance de la presse et d’une “opinion publique” en Chine moderne. Dans une démarche d’histoire sociale, mes centres d'intérêt englobent aussi l'histoire spatiale, les outils et méthodes numériques et l'épistémologie des sciences sociales en général.

Vous aimerez aussi...

Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search