Keio University

[Feature: Fukuzawa Yukichi and Statistics] Roundtable Discussion: Looking at "Statistics" over a 150-Year Span

Publish: June 05, 2020

Participant Profile

  • Hiroshi Saigo

    Professor, Faculty of Political Science and Economics, Waseda University

    Graduated from the Faculty of Economics, Waseda University in 1984. Withdrew from the Doctoral Programs at the Graduate School of Economics, Waseda University in 1992. Appointed as a full-time lecturer at the Faculty of Economics, Waseda University in the same year. Has held his current position since 1999. Specializes in statistical survey theory and the history of statistical systems. Author of "Introductory Statistical Analysis" and other works.

    Hiroshi Saigo

    Professor, Faculty of Political Science and Economics, Waseda University

    Graduated from the Faculty of Economics, Waseda University in 1984. Withdrew from the Doctoral Programs at the Graduate School of Economics, Waseda University in 1992. Appointed as a full-time lecturer at the Faculty of Economics, Waseda University in the same year. Has held his current position since 1999. Specializes in statistical survey theory and the history of statistical systems. Author of "Introductory Statistical Analysis" and other works.

  • Hiroe Tsubaki

    Other : Director, The Institute of Statistical Mathematics, Research Organization of Information and Systems (Professor Emeritus)Other : Special Juku Member

    Completed the Master's program in Mathematical Engineering and Information Physics at the Graduate School of Engineering, The University of Tokyo. Doctor of Engineering. Specializes in applied statistics. Served as a full-time lecturer at the Faculty of Science and Technology, Keio University from 1987 to 1997. Has also taught statistics at SFC since 1995. After serving as President of the National Statistics Center, he has held his current position since 2019. Professor Emeritus at the University of Tsukuba.

    Hiroe Tsubaki

    Other : Director, The Institute of Statistical Mathematics, Research Organization of Information and Systems (Professor Emeritus)Other : Special Juku Member

    Completed the Master's program in Mathematical Engineering and Information Physics at the Graduate School of Engineering, The University of Tokyo. Doctor of Engineering. Specializes in applied statistics. Served as a full-time lecturer at the Faculty of Science and Technology, Keio University from 1987 to 1997. Has also taught statistics at SFC since 1995. After serving as President of the National Statistics Center, he has held his current position since 2019. Professor Emeritus at the University of Tsukuba.

  • Takeharu Okubo

    Faculty of Law Professor

    Keio University alumni (1995, Politics). Withdrew from the Doctoral Programs at the Graduate School of Social Sciences, Tokyo Metropolitan University in 2000 after completing the required credits. Ph.D. (Political Science). After serving as an Associate Professor at Meiji University, he has held his current position since 2019. Specializes in the history of Oriental political thought and comparative political thought. Author of "Political Concepts of Modern Japan and the Netherlands" and other works. Member of the Keio University Fukuzawa Memorial Center for Modern Japanese Studies.

    Takeharu Okubo

    Faculty of Law Professor

    Keio University alumni (1995, Politics). Withdrew from the Doctoral Programs at the Graduate School of Social Sciences, Tokyo Metropolitan University in 2000 after completing the required credits. Ph.D. (Political Science). After serving as an Associate Professor at Meiji University, he has held his current position since 2019. Specializes in the history of Oriental political thought and comparative political thought. Author of "Political Concepts of Modern Japan and the Netherlands" and other works. Member of the Keio University Fukuzawa Memorial Center for Modern Japanese Studies.

  • Kunihiro Baba (Moderator)

    Affiliated Schools Teacher, Keio Shonan Fujisawa Junior and Senior High School

    Keio University alumni (1992, Science and Technology; 2005, Ph.D. in Science and Technology). Ph.D. in Science. Has held his current position since 1994. Specializes in mathematical statistics and data science education. Also teaches statistics at the two faculties of SFC as a part-time lecturer. Co-author of "Statistics as Problem-Solving Science." Member of the Keio University Fukuzawa Memorial Center for Modern Japanese Studies.

    Kunihiro Baba (Moderator)

    Affiliated Schools Teacher, Keio Shonan Fujisawa Junior and Senior High School

    Keio University alumni (1992, Science and Technology; 2005, Ph.D. in Science and Technology). Ph.D. in Science. Has held his current position since 1994. Specializes in mathematical statistics and data science education. Also teaches statistics at the two faculties of SFC as a part-time lecturer. Co-author of "Statistics as Problem-Solving Science." Member of the Keio University Fukuzawa Memorial Center for Modern Japanese Studies.

画像:「万国政表」(1860[万延元]年)

現在の統計学ブーム

馬場

今日は「福澤諭吉と統計学」というテーマの特集座談会ということで、統計学を150年のスパンでご出席の専門家の方と見ていきたいと思います。

現代はビッグデータの時代、またAIの時代と言われていますが、ビッグデータから意味のある知恵を引き出すのが統計学になります。また、AIの基盤となっているのもデータサイエンス、統計学ですから、現在、統計学ブーム、データサイエンスブームが起こっているわけです。

まず、今のこの状況をどのように見られているか伺いたいと思います。椿さんからいかがでしょうか。

椿

私は統計数理研究所の所長をしています。統計数理研究所は76年前、戦争中の1944(昭和19)年にできました。当時の新聞を見ると「敵米国では今次の戦争を数学戦または物理戦と称し、統計数学者の組織的動員を行っており」とあります。それから75年たったわけですが、戦争ではありませんが、データを用いて、様々な政策や行動、または何らかの運用が直接導かれる機会が多くなりました。データが非常に重要で、価値が高い時代になったと思います。そのために、統計教育やデータサイエンス教育が小中高から大学に至るまで充実を求められています。

そのような時代背景の中で統計数理研究所は研究機関ですが、日本の大学において、データサイエンス学部が、まだできたばかりということもあり、当面はそういった高等教育の支援をしていかなければいけないということになっています。

西郷

私自身が統計学と初めて出会ったのは大学1年生の時です。最初のうちは、いわゆる経済に応用される計量経済学に興味を持っていました。そこからだんだん、統計調査をはじめとした統計のデータをどのように作るべきなのかということに興味を持ち始め、最近ではいわゆる統計学、あるいは統計調査というものが日本にどのように移植されてきたのかというところに興味を持つようになりました。

しかし、その部分に関しては最近興味を持ち始めたばかりですので、今日は皆さんから教えていただければと思います。最近のデータサイエンスというのは従来の統計調査とはまた別の形で捉えているデータなので、ここをどのように自分のこれまでの研究から考えていくかということが課題です。

大久保

私は東洋政治思想史・比較政治思想史を専門としています。主な関心は、統計及び統計学を政治思想史の観点から検討することです。

近代国家の成り立ちを考えるとき、統計がどのような意味を持つか。この思想課題について、西洋世界ではミシェル・フーコー、イアン・ハッキング、T・M・ポーターらによる研究蓄積がありますが、近代日本はどうであったのか。幕末明治の転換期、日本の人々は西洋の統計知とどのようにして出会い、それは近代日本の国家形成にいかなる役割を果たしたのか、その出発点となる近世蘭学まで遡りながら研究しています。

昨年、厚生労働省から毎月出される「勤労統計」の不適切調査問題がありました。これもまた現代日本の国家統治のうちに統計をどう位置付けるかという本質的な問題に関わっており、明治維新から150年たった今日、改めて問い直すべき課題であると考えます。

馬場

私は中学・高校の教員をしていますので、統計教育の立場から少しお話をしますと、この10年くらいで統計教育が、飛躍的と言っていいほど非常に充実してきたと考えています。

昔から統計というのは中高の数学科の中にあったのですが、長い間、実際には教えられてこなかった。しかし、前回の学習指導要領の改訂で統計学がやっと本格的に導入され、今年度の改訂では小学校から高校まで、数学だけではなく情報やいろいろな教科で統計を扱うことになっています。いかに統計教育を充実させていくのかということが喫緊の課題かと思います。

日本の大学にはこれまで統計学部というものがなくて、3年前にようやく滋賀大学にデータサイエンス学部が開設されました。ですけれども、慶應のSFCでは、20年以上前からデータサイエンス教育に力を入れてきています。

早稲田も2017年からデータ科学総合研究教育センターを創立されていて、今年の4月から、「データ科学センター」に改組されたようですね。

西郷

はい。そのセンター設立に先立って、統計教育を全学的に展開したいと考えている教員は結構いたのです。2010年に統計関連学会連合大会が早稲田大学であったのですが、私自身実行委員の1人で、統計関連の学会に所属している人たち全員に一度集まってもらうと、学内に統計学の先生方が15、6人おられ、こんなにたくさんいらっしゃることに驚きました。その後、現在の田中愛治総長も政治学の中に統計分析を早くから取り入れられている方ですので、田中総長と意気投合した形で、センター長の松嶋敏泰先生が全学を統合するような形の統計組織を作ったのです。

まだできたばかりですが、これが定着すれば、そのセンターが中心になって全学的な統計教育、ひいてはデータサイエンス教育に結び付くのではないかと思っております。

150年前の「統計熱」

馬場

最後にまた現代に戻ってきたいと思いますが、幕末から明治にかけての日本で統計学ブームがあったということです。まず、大久保さんからそのあたりのお話をしていただけますでしょうか

大久保

慶應義塾大学名誉教授であった速水融先生も指摘されたように、幕末明治初期の日本では「統計熱」と言われる現象がありました。その先鞭をつけたのが福澤諭吉による『万国政表』(福澤校閲、岡本博卿訳)です。同書は、オランダ人プ・ア・デ・ヨング(P.A.de Jong)が記した『地球上の全ての国々についての統計表』を福澤たちが翻訳した作品で、1860(万延元)年に出版されました。

むろん福澤以外にも、後に日本近代統計の祖と言われる杉亨二や加藤弘之、西村茂樹、箕作麟祥ら多くの蘭学者・洋学者が同時期に西洋の統計に関心をもち、翻訳に取り組みました。

そのなかでも1つの画期となるのが、西周と津田真道のオランダ留学です。1863(文久3)年にオランダに渡った西と津田は、彼の地で経済学と統計学を専門とするライデン大学教授シモン・フィッセリングから直接、ヨーロッパ統計学を学び、その成果を日本に先駆的に導入しました。

馬場

17世紀にヨーロッパで統計学が始められ、非常に画期的な学問であるということで、ヨーロッパでまず統計学ブームが起こっていました。日本でも統計学に対して西洋と同じように思われたということでしょうか。

大久保

東アジアでは、『書経』や『日本書紀』にみられるように、古くから土地・人口調査が行われ、また徳川政治体制下では、「人別改(にんべつあらため)」が実施されていました。

しかし幕末明治初期の学者たちは、統計学を、それとは異なる未知の学問、「鉄道電信蒸気」とならぶ「文明世界の新学術」と捉えました。そのため、森鷗外をも巻き込んで、“statistics” を何と訳すかという訳語論争まで生まれました。

このことは、同時代ヨーロッパの統計学をめぐる動向とも連関します。ゲーテが『イタリア紀行』で「今日のような統計ばやりの時代」と記したように、18世紀末から19世紀のヨーロッパでは、巷に統計的な資料があふれ、社会の統計化が進められました。そこから、ケトレーらを中心に、新しい科学として近代統計学が形成されます。それはまたヨーロッパの近代国家形成とも深く関わっていました。

そうした真っ只中にまさに日本は開国を迎え、「文明世界の新学術」として、統計学に触れたのです。

椿

私は、どちらかというと確率を主体とする近代統計を学んでいますが、ドイツ国勢学派というか、明治時代に影響を与え、今の調査統計につながる流れも、ケトレーのところで理系的な統計と文系的な統計の両方が融合し、一挙に近代的科学として確立したのだと思います。ケトレーの方法が大きな影響を与え、それがイギリスにわたって、今の数理統計に近いものになりました。

『文明論之概略』に見られる統計的思考

馬場

福澤と統計とのかかわりをより詳しく見ていきたいと思います。『万国政表』では統計データを訳して出版したわけですが、統計的な考え方というものの有用性に、いち早く気付いた1人という面もあるのではないかと思うのです。

大久保

その通りだと思います。先に触れた幕末期の『万国政表』は、各国の国情をデータで示した統計表です。しかし福澤はそこからさらに思索を深め、「統計的な思考とは何か」という課題に踏み込んでいきます。

その1つの成果が、『文明論之概略』(1875[明治8]年)です。良く知られるように、同書で福澤はバックルの『英国文明史』に取り組んでいます。バックルがこの作品で文明史の方法論として導入しているのが、ケトレー統計学です。

バックルはケトレーの統計学により、歴史学の科学化が可能になったと指摘します。バックル『英国文明史』の主題は、膨大な資料を収集し、大量観察を通じて人間社会の精神現象のうちから一定の数量的規則性を導き出す統計学の手法を用いることで、科学的な文明史の叙述を試みることでした。

福澤は『文明論之概略』のなかで、この『英国文明史』を媒介に、犯罪率や死亡率をめぐるケトレー統計学の成果を紹介し、次のように指摘します。「スタチスチク」こそ、「天下衆人の精神発達」を観察・比較し「真の情実を明にする」、「文明を論じる学者」の方法論である、と。

このあたり、福澤はやはり思想家として非常に優れた嗅覚の持ち主であったと考えられます。

馬場

福澤はおそらく自然科学という分野は非常に法則的な世界であるけれど、自然科学だけではなく、社会科学においても、統計的なものの見方をすれば、同じように法則性を見つけられることに気付いたのでしょうね。

それで非常に興奮して、これを使えば社会的な問題も捉えられる、と思った。そこにいち早く気付いて著作の中で紹介したという点が、やはり福澤の先見性だったのではないかと思います。

大久保

興味深いことに、ケトレーはもともと天文学を研究し、そこから統計学に進みました。福澤が学問的基礎とした江戸時代の蘭学も、宇宙の法則をめぐる天文学の分厚い伝統と蓄積を有していました。

「天を測る」天文学から「人間社会を測る」統計学へ。福澤の統計学的営為は、近世蘭学の文化的鉱脈の延長線上に位置づけることも可能ではないかと考えます。

様々な統計学の側面を導入

馬場

福澤は同じ『文明論之概略』の中で、大数法則だけではなくて、婚姻の数が米相場の値で変わるのだという、因果を導くような点も統計学の1つの側面として話しています。また、『時事小言』(1881[明治14]年)の中では、議論としてはかなり粗いですが、士族をバックアップするという趣旨で、正規分布表なども導入して「天才の家系」を紹介しています。

ですから、大数法則だけではなく、いろいろな統計の考え方もかなり読み取って、それを著作の中に入れていたのではないかと思うのです。

椿

欧州における天才の家系の研究というのはダーウィンの従兄のフランシス・ゴルトンがやったんですね。『遺伝的天才』(Hereditary Genius、1869)という著作の中で初めて「スタティスカル・サイエンス」を作らなければいけないということで、相関概念とか分位点の概念を提案してくる。

この流れの中には、ゴルトンの強い影響を受けたカール・ピアソンの、「グラマー・オブ・サイエンス」(『科学の文法』、The Grammar of Science、1892)という概念があります。これは「自然科学のように対象が科学的であるから科学なのではなくて、プロセスが科学を科学とする。どんなものでも科学にできる」ということを前提にしている考えです。

しかし、今のお話を伺うと、福澤先生はケトレーから直結しているかもしれませんし、福澤先生のやっていらっしゃる時期のほうが早く感じられる部分もあります。

大久保

『時事小言』の「天賦の才能」と遺伝に関する正規分布は、福澤自身、実際にゴルトンを読んで紹介したものです。その議論を援用しながら、武士の家系の問題を考察しています。

椿

そうなんですね。

大久保

ここからは福澤が一貫して統計の問題に関心を持っていたことがうかがえます。

福澤は『文明論之概略』で、結婚を取り上げ、出雲大社の縁結びの神説など、縁談を運命のように考える人が多いが「スタチスチク」からみると、それは違う、その年の婚姻の多寡は「米相場」によって決まる、と喝破しています。福澤にとって統計学は、人々の間に流布する偏見や謬説をひっくり返す、最強の学問だったと思います。

それゆえに、ゴルトンを『時事小言』の中で紹介するときは、人間はみんな平等だというけれども、しかし実際には平等ではない、人間の才能は家系によって違いがある、と言うわけです。

ただし実はこれは、福澤の有名な『学問のすゝめ』(初編、1872年)の冒頭の一文、「天は人の上に人を造らず人の下に人を造らず」と矛盾します。

一方で冷徹に一般的事実を解明する統計学を重んじ、他方で人間の平等性など倫理にかかわる規範的な政治哲学を講じる、そうした両義性こそ、福澤諭吉の面白さだと思います。

「実学の精神」と統計

馬場

福澤の複眼思考というところかなと思います。椿さんがおっしゃったように、後にカール・ピアソンは「統計というのは科学の文法だ」と言うわけですが、福澤の1つの大きな思想として「実学」というものがあろうかと思います。

福澤は「実学」という漢字に「サイヤンス」とフリガナを振っているわけですが(「慶應義塾紀事」)、社会科学であっても科学的にものを見るというのが福澤の実学であって、ではどうしたら科学的に見ることができるのかといった時に、その1つの大きな道具が統計学だったのではないか。福澤は統計学を知ったがために、どんな分野においても科学的にものを見ることができるということに気付き、実学の精神につながっていったのではないかと思うのです。

大久保

同感です。福澤が東洋の学問に欠けているのは数理学だと訴えたことは良く知られています。実学を唱える福澤にとって、統計学は非常に重要な意味を持っていたと考えられます。

実際、『福翁百話』には、「統計全体の思想」なき人とはともに文明のことを語れない、という一文があります。個々の現象だけではなく社会全体を見渡し、事実を客観的に分析する統計学的思考が重要である、と。

同時にこれはまた、社会それ自体を1つの自律的な存在として捉えることでもあります。“society” の語をどう翻訳するかが明治初年に問題になりましたが、「統計全体の思想」は、国家とは異なる、人々が作り出す自律的な社会への眼差しでもあります。

馬場

カール・ピアソンは、やはり福澤と同じような意味合いで『科学の文法』という著書を書かれたということでしょうか。

椿

そうだと思います。イギリスでは、ナイチンゲールがものすごくケトレーを崇拝していて、ゴルトンに「オックスフォードに統計学科を作れ」と言ったわけです。ナイチンゲールとダーウィンの両方がゴルトンのいとこです。そういう意味では、ダーウィニズムの影響も受けたし、統計的問題解決を発想として使います。

時代の流れの中で、統計によって社会を改革するために、いろいろな社会現象をも法則化できる、その法則化プロセスを体系化したのが、物理学者だったカール・ピアソンです。ケトレーは天文学者でしたが、カール・ピアソンも力学や幾何学の先生でした。もともと極めて物理学的センスのある方だと思うのです。だから、数理的方法を使って、社会現象にかかわる体系を描けるようになった。そういう意味で、イギリスでは20世紀前半、近代統計学、数学的な統計学が非常に発達したのだと思います。

バックルの統計学的方法論

西郷

非常に興味深くお話を伺いました。福澤が捉えていた統計ないしは統計学は、因果関係を同定するものとしての統計学なのか、あるいは現象をあるがままに捉えるための統計学という意識だったのか。どちらのほうが強かったと考えていますか。

大久保

今日的な統計理論から見てどこまで整合性がある議論ができるかどうかはわかりませんが、基本的には両面あったと考えられます。例えば『時事小言』でも、第4編では各国の国情に関するデータを比較検討しています。

しかし第5編に入ると、先述のように、ゴルトンの天賦の才能と遺伝や家系をめぐる議論へと歩を進めています。

椿

ピアソンは因果関係をあまり哲学的に捉えずに、単純に物事の順序、といったものすごく記述的な立場ですね。それがいいのかどうかはまた別問題ですが。

大久保

この点に関連して指摘すべきは、明治初年のバックル問題だと思います。『英国文明史』で歴史学の科学化を志向したバックルは、ケトレーを超えて、より哲学的かつ因果論的に統計学の成果を捉え、それを自らの文明史へと援用しました。

特にバックルは、「統計的」手法とデータを用いて、非ヨーロッパ世界であるアジアにおいては、その気候や土壌から恐怖と迷信が蔓延(はびこ)り、専制政治が行われているため、文明の発展が妨げられていると分析しました。

しかし、もしバックルの唱える自然決定論的なアジア停滞論が事実であるならば、日本を含むアジア諸国は、いつまでも未開・半開にとどまり、文明化できない運命になってしまう。この問題は明治初年の学者たちを悩ませ、『明六雑誌』でも論争へと発展します。

福澤の『文明論之概略』は、一方でバックルが用いる統計学的手法を高く評価しながら、同時にバックルが唱えるオリエンタリズムに裏打ちされた自然決定論的かつ運命論的なアジア停滞論を乗り越え、独自の文明化構想を導き出すという、極めて困難な思想課題に挑んだ書物と言えます。

椿

面白いですね。

明六社の人々と統計学ブーム

馬場

ライデン大学から帰ってきた西周と津田真道、それから西村茂樹、杉亨二、皆、明六社ですね。このように明六社の人々が中心となって統計学が非常にブームになったと思います。福澤だけではなく、彼らが皆、統計というものの魅力に気付いて、取り組むわけですね。

西郷

やはり杉亨二が日本の、いわゆる公的統計ないしは政府統計を作っていく上では非常に大きな役割を果たしたと思います。

杉は、統計に興味を持ったということだけではなくて、それを絶対に実施すべきなのだと、統計のメーカーとして、つまり、統計を作る、そしてその作った統計を使って国政を進めていくべきなのだと最初に強く意識した人です。彼がいたお蔭で、今日の日本の産業統計等の統計、そして日本の統計組織が作られていったのだと思います。

大久保

杉亨二は徳川末期、西周や津田真道、加藤弘之らとともに、蕃書調所(ばんしょしらべしょ)に所属していました。蕃書調所は、徳川政権が西洋事情の調査と洋学教育を目的に設立した学問機関であり、多くの優れた蘭学者・洋学者が登用されました。杉は自伝で当時のことを次のように回顧しています。

杉によれば、彼はオランダの新聞『ロッテルダム・コーラント』を読み、ヨーロッパには統計というものがあると知って、面白いと思っていた。そこに西と津田がオランダ留学から帰ってきて、ライデン大学教授フィッセリングから学んだオランダ語の統計学講義ノートを見せます。杉はそれを読んで、一気に統計学に深入りしたと言います。

フィッセリングは日本では法学者として知られますが、実は当時のオランダを代表する統計学者・経済学者でした。隣国ベルギーでケトレーが統計学を大成させ、政府統計局を創設したことに影響を受け、オランダでも政治的に中立な政府統計局を作ろうという動きが高まります。フィッセリングは、その活動の中核にいました。

フィッセリングの統計学講義の講義ノートは、その後、1874(明治7)年に津田の手で翻訳されますが、杉亨二はすでに幕末期に、西と津田を通じて、原本のオランダ語原稿に触れていたのです。そうして、日本も統計局が必要であると考えるに至った杉は、明治期に入り、政表課の創設を主導するのです。

大隈重信と「スタチスチクの仲間」

馬場

大隈重信も「統計伯」と呼ばれるぐらい非常に統計と深い関わりがありました。

1879(明治12)年に福澤から大隈に宛てた書簡があります。その中で福澤は慶應義塾の塾員を13名、「スタチスチクの仲間」ということで紹介をする。それから、杉亨二や呉文聰(くれあやとし)など3名を「統計局の人」として推薦する。その書簡あたりが、おそらく福澤と大隈とが統計を介して非常に接近していった頃なのではないかと思います。

Okubo

One pipe connecting Fukuzawa and Okuma was the human exchange involving the founding of the Statistics Board (1881).

However, it should be noted that while Koji Sugi's name appears in Fukuzawa's letters, Sugi was already working hard at the Seihyo (Political Tables) Section of the Dajokan (Grand Council of State) to realize independent government statistics. Meanwhile, a "Statistics Bureau" was also established within the Ministry of Finance, where Okuma served as the Minister of Finance, and statistical administration began to diverge within the Meiji government.

In other words, while there was human exchange through statistics among them, there was also a tug-of-war and rivalry over statistical administration within the government behind the scenes.

A period was put to this conflict by the establishment of the Dajokan Statistics Board by Councilor Shigenobu Okuma in May 1881. As a result, Sugi's Seihyo Section was subsumed and dissolved into the Statistics Board.

At this time, the Keio University group, including Fumio Yano, Yukio Ozaki, Tsuyoshi Inukai, and Takuzo Ushiba, who were sent based on Fukuzawa's recommendation, worked as Okuma's right-hand men.

The background to this is thought to be more than just the purpose of enhancing government statistics; it was a political alliance and joint struggle between Okuma and Fukuzawa toward the early opening of the National Diet. This developed into the Political Crisis of 1881.

Tsubaki

In 1871, the Seihyo Section was created in the Dajokan, and Koji Sugi went there. On the other hand, at the Ministry of Finance, Hirobumi Ito was involved in statistics. It is possible that people like modern bureaucrats led that side.

Was the group from the Bansho Shirabesho (Institute for the Study of Barbarian Books) taking the initiative in the Seihyo Section?

Okubo

I believe so. Sugi and his colleagues thought it was necessary to establish a politically neutral central statistics bureau to form a modern state, and that the core of its role should be survey statistics centered on population surveys. In contrast, Okuma and others focused on creating operational statistics by collecting figures from various ministries. From Sugi's perspective, there was likely a strong awareness that the role of a statistics bureau should not be limited to that.

Tsubaki

At that time, the terms "Seihyo" (Political Tables) and "Tokei" (Statistics) were indeed separated, and Koji Sugi published "Nihon Seihyo" (Japan Political Tables) in response to Fukuzawa's "Bankoku Seihyo" (International Political Tables).

I often hear that Sugi was very particular about the word Seihyo, or "Stachisuchiku" (Statistics) itself, but it seems the word "Tokei" was actually used first by the Ministry of Finance.

Saigo

There is also the argument that the way the word "Tokei" was used is an issue. In the past, there was a verb form "Tokei-suru," which had a meaning similar to "to settle accounts," used when summarizing something. The Statistics Department or Statistics Bureau established in the Ministry of Finance used it in this sense. I feel that the meaning of the word "Tokei" was not necessarily the same as the "statistics" or "statistical science" we imagine today.

Therefore, Koji Sugi likely used "Seihyo" from the standpoint that it was different from the "Tokei" used until then. There are also two ways to write "Seihyo": one using the characters for "Political Tables" and another using the characters for "Table-making," the same as the current Seihyo (Tabulation) Section. Koji Sugi used both.

The Difference Between Okuma and Koji Sugi

Saigo

It is not very clear how Fukuzawa influenced the statistical organizations within the government. I have the impression that, at the very least, Okuma and Koji Sugi did not necessarily get along well within those organizations.

Regarding the establishment of the Statistics Board, Okuma asked a man named Shigemasa Aihara, who is thought to have been Sugi's subordinate, to visit a man named Mayet, who worked at the Prussian Bureau of Statistics, and drafted the "Draft Statistics Regulations" by imitating the Prussian Bureau of Statistics. Looking only at that, it seems as if the relationship between Okuma and Sugi was reasonably good.

However, as soon as the Statistics Board was established, Sugi likely expected to play a central role as an expert, but instead, people like Yukio Ozaki and Tsuyoshi Inukai, who had little to do with statistics until then, ended up occupying the Statistics Board. Despite being a statistics expert himself, Sugi was not given an important position within the Statistics Board.

Perhaps that is also why, when the Statistics Board was reorganized into the "Cabinet Statistics Bureau" in 1885, key figures like Sugi resigned.

Okubo

In fact, the Keio faction surrounding Okuma—Yano, Ozaki, and Inukai—recalled that their true purpose in serving at the Statistics Board was to prepare for the early opening of the National Diet. They said Okuma's intention was to select talent from the private sector, bring them into the government, and have them practice administrative affairs, because once the Diet opened, many government delegates would be needed to explain state affairs.

Naturally, the Seihyo Section group under Sugi seems to have resented these moves. Sugi's disciple, Katsura Go, stated that regarding Yano, Ozaki, and Inukai, there were "many people who came from journalism and didn't understand what statistics were," and that for those like themselves who "considered statistics their life's work," it was "extremely disappointing."

It is quite difficult to determine what position Fukuzawa held and what role he played in this. Since he recommended graduates of Keio University, he must have had some involvement, but Fukuzawa's presence is hidden in the shadows and does not appear on the surface. This is one of the difficult problems when considering the Political Crisis of 1881.

Evaluation of Shigenobu Okuma

Baba

Okuma may have had political motives around 1881, but in later years, in 1898, he said something to the effect that "statistics are extremely important in conducting politics. To advance national administration, the government should take proper statistics and conduct politics based on them." Okuma, as a politician, possessed ideas that lead to what is commonly called EBPM (Evidence-based Policy Making) today.

Looking at Okuma's long life as a politician, I think he can be evaluated as someone who, unlike Fukuzawa, highly recognized the importance of government and official statistics and put considerable effort into their development.

Saigo

I mentioned earlier that Koji Sugi was treated coldly by Okuma. However, while the so-called Okuma-Itagaki Cabinet (1898) was very short-lived and Japanese history textbooks suggest it did nothing, Okuma did something very significant in terms of statistical administration.

That was upgrading the Cabinet Statistics Section, which had been downgraded to a section during Hirobumi Ito's cabinet, back to a Statistics Bureau. That Statistics Bureau has continued to exist as a "Bureau" through World War II to this day. This is a very large contribution Okuma made to government statistics, especially to the Statistics Bureau. While there may have been some political intent, I believe the effect continues to this day.

Tsubaki

Okuma considered the Statistics Board to be on the level of a ministry or agency in today's terms, didn't he? When I read Okuma's "Proposal for the Establishment of the Statistics Board," which I've often been shown at the Statistics Bureau, I think he was very conscious of something like what we now call EBPM.

Saigo

That's true. However, Okuma may have had a stronger awareness as a user of statistics rather than a maker. Therefore, he might have thought that statistics were something that could be created simply by summarizing operational records.

Okubo

Okuma himself advocated that only after the government creates "perfect statistical tables" can it know the "current state of the nation" and judge the "merits and demerits of policies."

An interesting point related to that is the insurance system. Yukio Fukuzawa pioneered the introduction of Western insurance systems to Japan through works like "Guide to travel in the western world." On the other hand, it was Eiichi Shibusawa who founded Japan's first insurance company, Tokyo Marine Insurance Company. According to Shibusawa's recollections, Shibusawa and Fukuzawa played shogi at Shigenobu Okuma's house, and during the match, the three of them discussed the insurance system.

Needless to say, the insurance system is established and takes root based on statistical thinking. Fukuzawa, Shibusawa, and Okuma. The fact that they looked at the state of society with a statistical gaze beyond their individual positions and worked to establish new systems such as insurance companies, banks, and the Statistics Bureau is extremely significant when considering the formation of modern Japan.

Katsura Go and the Culture from the Bansho Shirabesho

Baba

Next, I would like to move on to the story of Fukuzawa's disciples and those around him.

As mentioned earlier regarding insurance, Taizo Abe, a Keio University alumni, founded Meiji Life Insurance. In this way, Fukuzawa also had his disciples work in fields that utilized statistics. Furthermore, when the Keio college was established in 1890, a "Statistics" course was set up in the department of political economy from the beginning, with figures like Katsura Go and Masao Yokoyama serving as lecturers.

Masao Yokoyama was a graduate of the Kyoritsu Statistics School founded by Sugi and was a person who worked hard to popularize statistics at the Statistics Company and elsewhere. In that respect, I think there were many people deeply involved in statistics around Fukuzawa and Keio University.

Tsubaki

I am also aware that Katsura Go, along with Koji Sugi, had a great influence on the census, but the source connecting to the people of the Meirokusha was the Bansho Shirabesho, which had a very large influence.

Katsura Go was the grandson of Genpo Mitsukuri, who was the head professor of the Bansho Shirabesho. In that sense, the culture there likely had a major impact. And I feel that the discussion of how to use statistical methods in society, which we have been talking about, existed there continuously, leading to the emergence of very talented statisticians in the early Meiji period. I believe Go was a disciple of Yukichi Fukuzawa and at the same time was exposed to the traditional culture of the Shogunate's Institute of Western Studies.

Baba

I heard that you, Mr. Tsubaki, are re-evaluating Katsura Go's side as a theorist.

Tsubaki

Seven or eight years ago, I had the opportunity to work with suicide researchers to create Japan's "suicide statistics." I was told that suicide research was a taboo in Japan and there was almost no research on "suicide statistics" itself. However, when I looked into it, there was a paper titled exactly "Suicide Statistics" among Katsura Go's papers.

It appeared in a work called "Statistical True Stories" in 1899, where he classified suicides, considered cultural influences, and compared suicides across countries. Moreover, he created statistical tables for suicides in Japan over a five-year period from 1890, categorized by gender, age/class, and 17 causes.

I was very impressed by this and read Go's book "Theoretical Statistics." I had the impression that this book best summarizes what the pioneers of the early Meiji period in Japan learned in order to utilize statistics in society.

It is well-constructed and interesting even to read today, covering which phenomena should be approached statistically, and how, when thinking about statistics, one must classify and consider causal systems that act on the entire population and those specific to a part of the group.

At the 100th Anniversary of the Census

Baba

Speaking of Katsura Go, 2020 is the year of the 21st census. Since the first one was in 1920, this year marks exactly the "100th anniversary of the census." The first one was during Prime Minister Takashi Hara's time. Both Koji Sugi and Katsura Go passed away without seeing its realization, but it was Sugi and Go who worked very hard, insisting that a census must be conducted. Katsura Go went to the United States to inspect the reality of the census and worked hard for its realization.

That attitude also connects to Okuma, whom we mentioned earlier. I think the idea is that the reason a census must be conducted is that everything begins with properly taking statistics and grasping the actual situation of the country.

As mentioned at the beginning, unfortunately, 100 years after the census, incidents have recently occurred where the importance of official statistics is being downplayed. In modern times, I feel we should look back once more at the passion of Go and Sugi for the census and use it as an opportunity to recognize the importance of official statistics.

Mr. Saigo, from your position involved in official statistics, what are your thoughts?

Saigo

I think there is no doubt that the creation of official statistics cannot proceed in the same way as before.

In other words, until now, taking statistics meant that the state did everything. At least in terms of official and government statistics so far, there has been an aspect of proceeding from a "top-down perspective," where the rulers take the statistics.

However, currently, it is becoming difficult to create various forms of official statistics through the method of "rulers taking statistics" alone. That is, the cooperation of those providing the data is becoming difficult in that form.

On the other hand, the types and structures of required statistics are becoming increasingly complex in line with scientific progress. In such an environment, if we are to continue creating official statistics in the future, the creators must adapt to the reality of needs—for example, by making internet responses possible for the census or creating accurate statistics while protecting privacy.

Okubo

What I noticed is the word "Kokusei" (National State/Census). In the original term "Population census," the word "country" or "state" does not appear anywhere. According to Professor Masahiro Sato's book "The Census and Modern Japan," when the census was implemented in 1920, statisticians at the time deliberately used the word "Kokusei"—which evokes the state of the nation, its power, and national wealth—to persuade politicians and secure a national budget.

And this word "Kokusei" had been used since the early Meiji period as a translation for "Stachisuchiku" (Statistics), along with "Hyoki," "Seihyo," and "Tokei."

However, as Mr. Saigo just mentioned, official statistics have reached a turning point. For 100 years, and even since the starting point of modern state formation 150 years ago, the "census" has had an aspect as knowledge related to the "situation" and "power" of the "state." If that state of the nation, the shape of the country itself, is reaching a turning point, how should we think about it from now on?

On the other hand, if we return to the starting point once more, the Meiji pioneers, including Koji Sugi, strongly argued that to create a new state, it was necessary to create perfect statistical tables, conduct surveys fairly from a politically neutral standpoint, and widely publish the results to the public. They all possessed high ideals and passion, appealing that statistics are the key to building a modern state.

Today, when the reliability of Japanese government statistics is wavering, the words of those who built the foundation of the modern state resonate deeply.

Tsubaki

For the Meiji pioneers to create this much statistics must have required an enormous effort to collect data compared to now. Moreover, I believe Sugi and Go were thinking about policies—what to do with that data.

100 years after the census, in an era of data science and big data, we need to clarify once again what exactly can be done with this data. We need to make clear how we use the census and official statistics, and what policies we derive from them. Conversely, I believe this is a chance to further strengthen the wisdom that people since the Meirokusha era possessed.

Natsume Soseki and Statistics

Baba

Mr. Tsubaki, you are also knowledgeable about the connection between Natsume Soseki and statistics. Could you speak a little about that?

Tsubaki

Of course, Soseki was not a statistician or anything of the sort, but he read Karl Pearson's "The Grammar of Science" thoroughly and, during his time studying in London, resolved to "turn literary criticism into a science."

Looking at his notebooks from his London days, there are various amazing things written, such as "the usefulness of a hypothesis is proportional to the probability that the hypothesis is valid," and the word "data" appears many times.

He believed that if one accumulated proper facts, the design of literature or literary criticism itself would become a kind of science. Moreover, he also criticized Karl Pearson, pointing out that while Pearson created a wonderful system from the standpoint of recognizing facts, if one considers this in terms of art, the purpose of art is not necessarily to establish order and reduce noise.

I think Soseki was a person who gave a very interesting influence to Japan just by introducing "The Grammar of Science" to Japan and to his disciples like Terada Torahiko.

Baba

That is likely a manifestation of the charm of statistical science, where various people, whether politicians or literary figures, noticed the appeal that statistics hold.

Tsubaki

I suppose so. Soseki was from a slightly later era than the predecessors mentioned earlier, but I feel that passion very much.

The Growing Need for Statistical Education

Baba

I believe we must not let this current statistics boom end as just a boom. I would like to hear your thoughts on the future of statistics, statistical science, and data science.

Saigo

I mentioned earlier that it has become impossible to take government statistics in the same way as before, but conversely, from the standpoint of users who use statistics, the development of technology is remarkable. For example, the level of the papers I wrote as a student is completely different from the papers students are writing now. One factor is that the available data has increased, and the available technology is completely different from the past.

For example, when I was an undergraduate student, it was almost impossible to perform a statistical analysis of musical pieces, but today's students can process audio so it can be analyzed as data, use image processing, and perform the analysis they want to do in about six months. So we have entered an era where anyone can analyze quite difficult data. However, I feel that not many people have fully recognized this yet.

If we can recognize that data, or statistics, are extremely useful for our lives, we will recognize that data is very important for the government to decide the state of national administration, for example. By increasing such awareness, it might become easier to collect the data that the country needs, or that we as citizens need. I hope we can achieve this in the field of statistical education, including at universities.

Baba

In terms of statistical education, the "AI Strategy 2019" announced by the Cabinet Office aims to have all university students, regardless of whether they are in the humanities or sciences, acquire a standard curriculum in data science by 2025.

In the coming era, EBPM and data-based, evidence-based discussions will be necessary in any field. Also, within massive amounts of big data, it is possible to lead to wrong conclusions if one intends to, so the ability to see through that and correctly interpret things is required.

This connects to Fukuzawa in terms of a scientific way of looking at things. It is not enough for a few statisticians to handle statistics as in the past; it is important for all citizens to have a statistical perspective. For that reason, we must put even more effort into statistical education.

Re-examining "Statistics" in Modern Japan

Okubo

Since I examine statistical issues from the perspective of the history of political thought, I would definitely like students studying statistical science today to look at it from a historical perspective at least once. Through that, I hope they will re-recognize the importance of statistical science while simultaneously cultivating a perspective that questions its self-evidence—asking if statistical science is truly omnipotent.

As we have discussed today, the foundation of current statistical science was built in the 19th century and was deeply connected to the formation of modernity. Therefore, at that time, both in the West and in Japan, there existed the fundamental question: "What is statistics?"

We mentioned Natsume Soseki earlier, but in terms of literature, Dostoevsky states the following in "Notes from Underground": "As far as I know, you," in representing human interests and happiness, "take average values from statistical tables and economic formulas." Anyone who goes against that is called an "obstinate anti-enlightenment theorist." "This, gentlemen, is our 19th century, in which Buckle also lived."

The 19th century, symbolized by Buckle, which uses statistical tables to show happiness. Dostoevsky's critique from the side of a literary figure asks whether that is true happiness.

This overlaps with the problem Fukuzawa faced in "An Outline of a Theory of Civilization." Can Asia truly escape from the stagnation theory of natural determinism? What is true civilization? While valuing "Stachisuchiku," Fukuzawa also touched upon the problems beyond it and pondered them.

In today's world where data science and AI technology are highly developed, is looking at the ideological issues that existed at these historical starting points merely the nonsense of an "obstinate anti-enlightenment theorist"? This is a question I would definitely like young people who will study statistical science to consider at least once.

Tsubaki

Currently, statistics has become a cross-disciplinary liberal art, a typical example of a scientific method, and we are in an era where it should be properly acquired between elementary school and higher education.

Until now, data was a prize beyond reach, and taking data itself was a huge cost. In such an era, it was difficult for general people to create statistics, but we have entered an era where that is possible, and the tools to analyze it have also advanced greatly. Recently, there are even circles where high school students study machine learning. In that sense, I think it has become a very good era.

However, while it is a cross-disciplinary liberal art, there is another question: "What does statistics serve?" For some, it will be policy. Furthermore, statistics contributes to various academic fields such as medical sciences. It can also be a tool to advance the expertise and knowledge of the user. However, conversely, there is a fear that statistics could dominate humans and take the form of "do it with this conclusion."

Since expectations for AI and statistics are currently too high, I believe we must once again firmly establish that they are merely tools to advance our knowledge and a cross-disciplinary liberal art.

Baba

Today, we looked back 150 years from the modern statistical science boom, and I was able to re-recognize that statistics has many different aspects. While noticing this, Fukuzawa and other people of the Meiji period tried to incorporate statistics and statistical science. I believe those efforts have led to the present day.

I believe statistical science and data science will play an increasingly important role in the world. As Mr. Okubo said, it is important not to forget the historical perspective and for many citizens to share a statistical perspective and the importance of statistics as part of their liberal arts education.

Thank you very much for today.

(Recorded online on April 16, 2020)

*Affiliations and titles are as of the time of publication.