卷:卷二《长安月下·关中六十夜》|主角:玄奘·Marco|地点:西安·大慈恩寺·大雁塔下·夜|典:《荀子·劝学》“驽马十驾,功在不舍”、《大唐西域记》、玄奘译场“证义·笔受·书手·润文”之制|主导感官:嗅(香、旧木、纸墨、夜露)|过所进度:空白·一字未刻|金融-国学对应:Turning One Man's Work into an Auditable Process(把一个人的活,变成一套可复核的流程)↔ 驽马十驾,功在不舍

段一·现身

那座塔,是唯一没有被重新盖过的东西。

第六十七夜。

马可从阿房宫回来,睡了四个小时,下午醒了,一直坐在窗前发愣。

他一直在看那条短信。

塬上有六户,愿意跟你坐下来说说。

他脑子里已经开始转了。六户,如果每户十二亩,就是七十二亩。第一年的垫资按每亩两千算,十四万四。加上一个小型气调库,不,不能自己建库,村里有库,租。

他打开电脑,新建了一个文件。

打了两行字,又删了。

删完他坐在那儿,忽然出了一身汗。

他刚才在做的事,是把六户人家,变成一份材料。

他合上电脑,出了门。


他不知道该去哪儿。天快黑了,他就顺着南门往南走,走到雁塔路,看见了那座塔。

大雁塔。

他前几夜在大唐不夜城见过它,那一晚它站在一片流光溢彩的仿唐建筑里,又老又暗。

今夜他从另一个方向来,从慈恩寺的正门进去。

寺里已经快关门了。游客散得差不多,只剩几个人在钟楼那边拍照。院子里有香的味道,那种很实在的、微微带甜的檀香,混着入夜以后的土腥气。

塔就在正中。

七层,方的,砖的,一层比一层收一点。没有雕梁画栋,没有彩绘,没有金顶。每一层的四面都开着券门,砖缝里长着草。

马可站在塔底下抬头看。

他忽然觉得这座塔有点像一样东西。

像一个书架。

他后来才知道自己为什么会有这个念头。

这座塔的形状不是为了好看。它一层一层往上收,底下最厚、最宽,往上越来越薄,那是因为承重。它每一层四面开门,是为了通风,砖砌的密闭空间会返潮,纸怕潮。

它的每一个设计,都在解决同一个问题:怎么让纸活得久一点。

“你看对了。”

一个声音在他身后。

马可回头。

一个僧人站在几步之外。

年纪很难说,他的脸上有很深的风霜痕迹,皮肤是那种在极强的日照下晒了很多年的颜色。可他的眼睛很亮,亮得不像一个老人。

他穿一件很旧的赤褐色僧袍,脚上是一双磨破了的麻鞋。

“大师。”马可躬身。

那僧人合掌:“玄奘。”


段二·古文镜像

“骐骥一跃,不能十步;驽马十驾,功在不舍。”
English Mirror: A thoroughbred, in a single leap, cannot cover ten paces; a poor horse harnessed for ten days gets there — the merit lies in not stopping.

“大师,”马可说,“这座塔,是您建的?”

“是我请建的。”玄奘说,“永徽三年。”

“为什么建?”

玄奘抬头看着那座塔。

“为了放东西。”

“放什么?”

“经。”玄奘说,“我从西边带回来六百五十七部。还有佛像、舍利。”

“放不下了?”

“寺里的屋子是木头的。”玄奘说,“长安城几十年就要烧一次。木屋子一烧,什么都没了。”

他望着那些砖:“我要一个不怕火的地方。”

马可站在那儿,忽然觉得心里有什么东西被拨动了一下。

一座塔。一千三百多年了。它不是为了纪念谁,不是为了好看,它是一个防火的库房

“大师,”他说,“我今天想问您一件事。”

“讲。”

“您西行,一共用了多少年?”

“出去五年,在那边求学游历十来年,回来一年多。”玄奘说,“前后十七年。”

马可算了一下。

“您出发那年多大?”

“二十七。”

“回来呢?”

“四十四。”

马可沉默了。

十七年。一个人从二十七到四十四。人一生最好的那十七年。

“大师,”他艰难地问,“您出发的时候,知道要走十七年吗?”

玄奘看了他一眼。

“我出发的时候,”他说,“不知道能不能活着走出凉州。”

“那您为什么还走?”

玄奘想了一下,答得很实在:

“因为在长安待着,也解决不了那个问题。”


段三·事件主体

那一夜他们在塔下的石阶上坐了很久。

寺里关了门。院子里只有几盏地灯,很暗。风把香炉里的余烟吹得斜过来。

“大师,”马可说,“我读过一些关于您的东西。可我一直有一个问题。”

“讲。”

“朝廷,当时是不许出关的,对吧?”

“对。”玄奘说,“贞观初年,突厥未平,朝廷禁百姓私自出蕃。我上表请求,三次,都不许。”

“那您怎么走的?”

玄奘的表情一点变化也没有。

私自走的。

马可怔住了。

“偷渡?”

“偷渡。”玄奘点头,“我从长安到秦州,到兰州,到凉州。凉州都督李大亮听说有个和尚要西行,派人来赶我回去。”

“那您……”

“我藏起来了。”玄奘说,“有两个小和尚,慧威法师派来的,他们带我夜里走,白天躲。”

“到了瓜州,我要过玉门关。守关的人查得很严。有个校尉王祥,他认出了我。”

“他抓您了吗?”

“他要抓的。”玄奘说,“他手上有凉州的访牒,就是通缉的文书,上面写着我的名字。”

马可听得很紧张。

“后来呢?”

“他问我为什么要走。”玄奘说,“我跟他讲了。讲了很久。”

“讲什么?”

“讲我为什么非去不可。”玄奘说,“我说,中原的经,译得不全,也译得乱。同一部经有好几个译本,彼此矛盾。我问过许多法师,谁也说不清。”

“我说:我要去看原本。

马可忍不住插了一句:“大师,那些译本,真的差别很大?”

“很大。”玄奘说,“同一句话,一个本子译成‘有’,另一个本子译成‘非有’。”

马可怔住了。

“这两个意思是反的。”

“正是反的。”玄奘点头,“可两个本子都传了几百年,各有各的师承,各有各的道理。”

他望着黑暗:“我问过好几位大德。他们的答复是,依你所受的那一家。”

“那您接受吗?”

“不接受。”玄奘说得很干脆,“一句话,不能有两个相反的意思。要么这个本子错了,要么那个本子错了。”

“要么,”他停了一下,“原文里根本不是这么说的。

他停了一下。

“王祥听完,把那张访牒,撕了。”


马可久久没有说话。

“大师,”他终于说,“所以您那一趟,是没有过所的。”

“一开始没有。”玄奘说。

“一开始?”

“我在西域走的时候,靠的是各国国王给的通关文书。高昌王麴文泰给我写过二十四封信,给沿路二十四个国家的国君,每一封信附一匹绫。”

“他为什么帮您?”

玄奘沉默了一会儿。

“他想留我。”他说,“他要我留在高昌,做他国中的法师。他说,高昌全国供养你。”

“您没答应。”

“我没答应。”玄奘说,“他生气了,说不放我走。”

“那您怎么办?”

“我不吃饭。”玄奘说,“三天。”

马可闭上了眼睛。

“第四天他哭了。”玄奘说,“他跟我约定:你去,学成了要回来,在我这儿住三年。我答应了。”

“他就给了您那二十四封信。”

“是。”

马可忽然问:“您后来回去了吗?”

玄奘望着黑暗中的塔。

“我回来的路上,专门绕道去高昌。”他说,“到了才知道,高昌国已经没了。”

“贞观十四年,唐灭高昌。麴文泰在唐军到之前,惊惧而死。”

香烟在夜风里散开。

“那三年,”玄奘说,“我没能还。”


马可坐在石阶上,心里堵得难受。

“大师,”他忽然问,“那玉门关外呢?八百里流沙,”

“莫贺延碛。”玄奘说,“八百余里,古称沙河。上无飞鸟,下无走兽,复无水草。”

“您怎么过的?”

“我迷路了。”玄奘说得很平常,“我找不到第四座烽火台。找了半天,找不到。我下马取水,皮囊重,一下没抱住,水全洒了。”

马可的手心开始冒汗。

“那……”

“我掉头往东走了十几里。”玄奘说。

马可怔住:“往东?回去?”

“回去。”玄奘点头,“我想着,回瓜州取水再来。”

“那您怎么又走了?”

玄奘看着他。

“我走了十几里,想起我在王祥面前说过的话。”

“什么话?”

“我说:宁可西行而死,不东归而生。”

他很平静地说:“我说过了。所以我掉头,又往西去了。”


“后来呢?”

“后来四夜五日,滴水未进。”玄奘说,“人和马都躺在沙上,不能起。”

“到第五夜半,起了风,凉了些。我勉强上马。那马不肯走我指的方向,它自己往一边跑。”

“跑了几里,有草。再走十步,有一个池子。”

他停了一下。

“那匹马,是我在瓜州换的一匹老瘦赤马。有个老胡人劝我换的,他说这匹马往来伊吾十五趟,认得路。”

“我当时嫌它瘦。”

马可久久没有说话。

风把香炉里的一点火星吹亮了一下,又暗了。

“大师,”他终于开口,“我今天来,本来想问的是另一件事。”

“讲。”

“我想问,怎么加快。”

玄奘看着他,没有说话。

“我在洛川那边,有六户人家愿意跟我坐下来谈。”马可说,“我很高兴。可我今天下午一坐到桌子前,我就开始算,怎么把六户变六十户,怎么把一年变半年。”

“我在算怎么加快。”

他抬起头:“大师,您用了十七年。我算过我自己剩下的时间,如果每个地方都要六十夜,我这一辈子最多做五十个地方。”

“可全国有几千个县。”

“所以我一直在想:有没有办法快一点。”

玄奘听完,站了起来。

“跟我来。”


他带马可绕到塔的北面。

那里有一块空地,现在种着几株松树。

“这地方,”玄奘说,“当年是我的译场。”

马可站住了。

“译场?”

“翻经的地方。”玄奘说,“贞观十九年我回来,太宗要我做官,我不肯。我求一件事:给我一处地方,给我人手,让我翻经。”

“他给了。弘福寺,后来是慈恩寺,最后是玉华宫。”

他望着那几株松树:“十九年。我在这几处地方,翻了七十五部,一千三百三十五卷。”

马可算了一下:“一年七十卷……”

“不是我一个人翻的。”玄奘说。

马可抬起头。

“年轻人,你以为翻经是什么样子?一个懂梵文的和尚,坐在灯下,一句一句地译?”

“……我是这么以为的。”

“那样译不了一千三百卷。”玄奘说,“那样一辈子译不完一部大经。”

他转过身,看着马可。

所以我没有那样译。


“我建了一个场。”玄奘说。

“什么叫场?”

“一整套人,一整套次序。”

他伸出手,一样一样地数:

译主。”他指了指自己,“我。我读梵本,口宣其义。”

证义。坐在我左边。他懂梵文,也通义理。我译错了,他当场驳我。”

证文。坐在我右边。他听我念的梵文原文,看有没有念错。”

书字。梵学僧。他按我念的梵音,写成汉字。”

笔受。他把梵音翻成汉文。”

缀文。他把翻出来的字句,按汉文的次序理顺。”

参译。他把译好的汉文,回头跟梵本核一遍。”

刊定。他把啰嗦的地方删掉,定字数。”

润文。他管文采。”

梵呗。他把定好的经文念出来,听顺不顺口。”

马可听得手心全是汗。

十个岗位。

“大师,”他艰难地开口,“这是,一条流水线。”

“我不懂你这个词。”玄奘说,“不过你说的意思,大概不错。”

“大师,”马可急着问,“这十个位置,同时有多少人?”

“译主一个。”玄奘说,“其余的,一个位置有时候一个人,有时候三五个人轮。译大经的时候,场里坐过二十几个。”

“一天译多少?”

“看经。”玄奘说,“难的一天几百字。顺的一天几千字。”

他停了一下:“可有一条规矩,从头到尾没破过。”

“什么规矩?”

当天译的,当天定。

马可怔住了:“为什么?”

“因为过了一夜,那些人对那一段的记忆就淡了。”玄奘说,“证义想不起来自己当时为什么点了头,参译也想不起来那句梵文的语气。”

“到那时候,你就只能对着字改字了。”

他望着渐亮的天:“对着字改字,改不出对的东西来。”


马可在那几株松树底下站了很久。

他忽然明白自己这七天在西安学到的所有东西,都汇到了这一刻。

杜甫说:你在成都那七个村靠的是你本人在场,那个带不走。 司马迁说:你替不了所有人跨那一级。 白居易说:他缺的不是钱,是输一次的余地。 王维说:靠位子的东西,随位子消失。 杜牧说:一个人能做的事,有极限。

而玄奘给了他答案的形状:

你不是要把一个人的活干得更快。

你是要把一个人的活,变成一件不靠这个人的事。

“大师,”他说,“我想问几个很具体的问题。”

“讲。”

“这十个位置里,哪一个最要紧?”

玄奘想都没想:

证义。

“为什么?”

“因为他是唯一一个,专门用来驳我的人。”玄奘说,“别的位置都在往前推。只有他往回拉。”

马可怔住了。

他忽然想起第五十七夜。那一夜他给自己安了一个否决权,他跟小周说:你在那个位置上的唯一职责,是替我说‘不’。

他当时觉得自己想得很妙。

原来一千三百年前就有人这么干了,而且干得比他严整得多。

“大师,那第二要紧的呢?”

参译。”玄奘说,“译完了,回头拿梵本核一遍。”

“为什么?”

“因为往前走的时候,人人都觉得自己是对的。”玄奘说,“只有回头核那一遍,才知道自己走偏了多少。”


“大师,”马可又问,“这十个人,是您挑的吗?”

“是。”

“怎么挑?”

“贞观十九年,我上奏,请征集天下的僧人。”玄奘说,“来了很多。我一个一个考。”

“考什么?”

“不考佛法。”玄奘说。

马可意外:“不考佛法?”

“佛法当然要通,可通佛法的人多。”玄奘说,“我考的是另一样。”

“什么?”

玄奘看着他,很慢地说:

我考他,肯不肯一辈子只做这十个位置里的一个。

马可怔在那里。

“年轻人,你想想:证文这个位置,一辈子就是坐在我右边,听我念梵文,看我念错了没有。他不署名,不出头,也不著书。”

“可要是没有他,我念错一个字,整部经就传错了。”

他望着黑暗:“肯做这个位置的人,很少。大部分来的人,都想做译主。”

“那些人您怎么办?”

“我请他们回去了。”玄奘说,“客客气气地请回去了。”

“一个都没留?”

“留过两个。”玄奘说,“我想着,人才难得,先放在别的位置上,慢慢磨。”

“后来呢?”

“后来那两个都出了事。”玄奘说,“一个在笔受的位置上,自己改我的意思,他觉得他那样译更好。”

“另一个呢?”

“另一个在证义的位置上,不驳我。”玄奘说,“他想着,将来要接我的位子,不能得罪我。”

他望着黑暗:“一个位置上坐了一个心在别处的人,这个位置就等于空着。

“可空着还好些。”他补了一句,“空着你知道它空。有人坐着,你以为它有人。”


天快亮的时候,马可问了最后一个问题。

“大师,那十七年,值得吗?”

玄奘没有马上回答。

他抬头看着那座砖塔。第一缕天光已经落在塔顶了,砖是暖褐色的。

“我给你换个算法。”他说。

“大师请讲。”

“我西行十七年,取回六百五十七部经。”

“我回来译了十九年,译出七十五部。”

他转过头:“年轻人,你听出问题了吗?”

马可算了一下,然后他忽然明白了。

“六百五十七部……您只译了七十五部。”

“对。”

“剩下的呢?”

“剩下的,我没来得及。”玄奘说,“麟德元年,我在玉华宫。正月里我译《大宝积经》,译了几行,把梵本放下了。”

“我跟弟子说:这部经的分量跟《大般若》一样,我的气力不够了。”

“那年二月我死了。六十三岁。”

院子里安静下来。

“所以,”玄奘说,“我这一辈子,走了十七年,译了十九年,一共三十六年。”

我干完了不到八分之一。

马可站在那儿,眼眶发热。

“那您……”

“年轻人,你刚才问我值不值得。”玄奘说,“你问这个话的时候,心里在算的是,我付出了十七年,得到了什么。”

“这个算法不对。”

“那该怎么算?”

玄奘伸手指了指那座塔。

你去看那些经,现在在哪儿。


马可后来查了。

玄奘译的《大般若经》六百卷,是他晚年在玉华宫译的,他译那部经的时候已经病重,弟子劝他删略,他不肯,说要全译。译完那年他就走了。

而那六百卷,一千三百年来,中日韩三国的寺院里,一直在念。

他译的《心经》,二百六十个字,今天有几亿人背得出。

他译的《瑜伽师地论》,那是他当年在长安问不明白、非要去印度看原本的那一部,成了整个东亚唯识学的根本典籍。

他干完了不到八分之一。而那八分之一,垫了一千三百年。

他还查到一件事。

玄奘死后,他的译场散了。他的弟子窥基、圆测各立门户,唯识学在东亚分了几支。剩下那五百多部没译的梵本,后来一部分残了,一部分不知所终。

可他建的那个“场”,没有散。

后来北宋的译经院,用的还是这一套:译主、证义、证文、笔受、缀文、参译、刊定、润文。名目略有增减,骨架一模一样。

他这一辈子最值钱的东西,不是那七十五部经。是那十个位置。


马可在本子上写:The Big Wild Goose Pagoda is a fireproof warehouse. Seven storeys of plain brick, no paint, no gilt, grass in the joints. Xuanzang had it built in 652 to hold 657 sutras he had carried back, because temple buildings in Chang'an are timber and the city burned every few decades. It is the only structure in this district nobody has rebuilt.

Seventeen years out, aged twenty-seven to forty-four. He left without permission — petitioned three times, refused three times, and went anyway. Wanted for illegal border crossing; the officer at Guazhou who held the warrant tore it up after hearing him explain why he had to go, and his reason was not devotional. It was that the existing Chinese translations were incomplete and contradicted each other, nobody he asked could resolve them, and he wanted to read the originals.

He lost his water bag in the Mohe Yanqi desert, turned back east for a dozen li, remembered a sentence he had said out loud to that officer, and turned west again. Four nights and five days without water. What saved him was an old thin red horse he had not wanted, which a Hu trader had told him knew the road, and which walked in a direction he had not chosen.

I came tonight to ask how to go faster. I said so plainly: six households in Luochuan, and by that afternoon I was already modelling how to turn six into sixty and one year into six months. He didn't argue. He walked me around to the north side of the pagoda and told me about the workshop.

Ten positions. Translator-in-chief; a verifier of meaning seated on his left whose job was to contradict him on the spot; a verifier of text on his right checking his Sanskrit pronunciation; a transcriber of sound; a receiver who rendered it into Chinese; an arranger for Chinese word order; a back-checker who took the finished Chinese and re-collated it against the Sanskrit; an editor to cut; a stylist; and a reciter to hear whether it read aloud.

Nineteen years, 75 works, 1,335 scrolls. Not one man with a lamp. He said flatly: that method cannot finish a single major sutra in a lifetime, so I did not use it.

The two positions he named most important: the man whose entire function was to contradict him, and the man who went back and re-collated the finished work against the source. His reason for the second is the best sentence I have heard this week — going forward, everyone believes they are right; only the pass back tells you how far you drifted.

And how he selected people: not on doctrine. He tested whether a man would spend his whole life occupying one of the ten seats. The text-verifier never signs anything, never publishes, and sits to the right listening for a mispronounced syllable — and without him a wrong character propagates forever. Most applicants wanted to be translator-in-chief. He sent them home politely.

Last thing, and it dismantled the question I came with. He brought back 657 works and translated 75. He put down the Sanskrit of the Ratnakūṭa a few lines in and told his disciples his strength was gone; he died a month later, sixty-three. Thirty-six years of work, less than an eighth of the material.

I had asked whether the seventeen years were worth it, and he told me the arithmetic was wrong — I was computing what he paid against what he received. His replacement: go and look at where those sutras are now. The Heart Sutra is 260 characters and hundreds of millions of people can recite it. The 600-scroll Perfection of Wisdom, which he refused to abridge while dying, is still chanted in three countries.

He finished under an eighth. The eighth has held for thirteen hundred years. — M.


段四·钩子结尾

天亮了。

寺里开始有声音,扫地的、开门的、远处早课的钟。

玄奘要走了。

“大师,”马可忽然说,“我能不能,再问一件事。”

“讲。”

“您那个场,十个位置,是一开始就想好的吗?”

玄奘摇头。

“不是。”他说,“一开始只有三个人。我,一个笔受,一个书手。”

“那后来呢?”

“后来出错了。”玄奘说,“译到第三部的时候,有人发现前面一部里有一处,我把两个梵文词混了。”

“那处译错了多少?”

“一句。”玄奘说,“可那一句,是那一段的关键。”

他望着渐亮的天:“那一次我改了七天。改完我坐下来想:要是没人发现呢?”

“所以就有了证义。”

“对。”玄奘说,“十个位置,是十九年里,一个一个补出来的。每一个位置,都是因为出过一次错。”

马可站在那儿,忽然全明白了。

他一直以为制度是设计出来的。

其实制度是,被错误一次一次逼出来的。

“大师,”他说,“那我现在只有六户人家。我该建几个位置?”

玄奘看了他一眼。

那一眼里有一种马可后来一直记得的东西,不是慈悲,是一个干过大活的人对另一个刚要开始的人的,实在的估量。

“一个。”他说。

“一个?”

“对。”玄奘点头,“你现在有六户,你自己一个人管得过来。你要是现在建十个位置,那十个位置就是空的,没有活给他们干,人就会去找活干,找出来的活全是废活。”

“那这一个位置是什么?”

玄奘往塔的方向走去,身影在晨光里开始淡。

记错。

马可怔住:“记错?”

“找一个人,或者就你自己,”玄奘说,“备一个本子。这本子上不记你做对了什么。”

“只记错的。”

“每一件错,记三样:错在哪儿,是谁发现的,为什么没有早点发现。”

他的声音已经很淡了:

记满一年,你的第二个位置,自己会从那个本子里长出来。

马可赶紧追问:“大师,那第三样‘为什么没有早点发现’,是最难的那一条吧?”

晨光里已经几乎看不见玄奘了。

可他的声音还在:

“是最难的。”

“也是唯一有用的。”

“前两样只让你知道错了。”

只有第三样,告诉你哪儿少了一双眼睛。


马可一个人站在大雁塔下。

太阳完全出来了。塔身被照成一种很暖的褐色,砖缝里的草看得清清楚楚。

他掏出那枚青铜过所。

铜面上还是空的。

他忽然不急了。

他掏出手机,给强磊回了一条短信:

强磊你好,谢谢你爸。

那句话是真的:不用现在给钱,三年以后再分。

不过我想先说清楚一件事:我这套办法在成都做了七个村,一百来户。在陕西,我一户也没做过。我不知道它在你们那儿灵不灵。

所以我不敢跟六户人家谈。

我想先跟一户谈。

十月我过来。谁都不用准备什么,我就想在你们家院子里,把去年那笔账,从头到尾,跟你爸一起算一遍。

发出去之后,他站在塔下,等了大概两分钟。

手机响了。

行。我爸说,那你十月来。

他还说,他早就想找个人把那笔账算清楚了。

他说他自己算了二十年,一直觉得哪儿不对,可他说不出来是哪儿。

马可捏着手机,站在一千三百年前那个防火库房的影子里,忽然笑了。

他抬头看那座塔。

七层砖,一层一层往上收。

底下最厚,往上越来越薄。

他忽然想起玄奘说的那句“六户只建一个位置”。

这座塔要是一开始就照第七层的厚度砌,它站不到第三层。

位置不是越多越好。是要跟你现在这个高度,配得上。


他忽然想起一件事,他在西安已经七天了,还有五十三夜。

而他这七天,什么也没做成。

被送出来一次,被拆掉一整套方法,在一个院子里蹲了一天,在一条谷里站了一夜,在一座没盖成的宫殿上待到天亮。

可他手上现在有一样东西,是七天前没有的:

一个十月要去的院子,和一本还没开始写的、只记错事的本子。


他往寺门口走。

走到门口,他忽然看见有人在等他。

不是杜甫。

是一个女人。

她穿着极其精致的唐式宫装,站得很直。年纪不好判断,二十几,也可能三十几。她的脸很美,可那不是让人想多看的那种美,是让人不敢多看的那种。

她手里拿着一支笔。

看见马可出来,她开口了。

第一句话就让马可站住了。

你写的那三十七页,我看过了。

马可怔住:“您是……”

那女人把笔在手里转了半圈。

“上官婉儿。”她说。

“我这一辈子的活,就是,替天下人,把文章分个高下。”

她看着马可,很平静地说:

你那三十七页,我给它评了个等。


欲知后事如何,请看下回,《婉儿称量天下士》

附录一·Dongpo's Note

Foreign friend — Xuanzang, and I will be brief, since I was never much of a stylist. That was somebody else's seat.

You came to ask how to move faster. I did not refuse the question; I answered a different one, which is what to do when you have understood that you personally do not scale.

I went west without permission. I asked three times and was refused three times and went anyway, and a border officer holding a warrant for my arrest tore it up after I explained myself — not out of piety, but because the reason was practical: our translations were incomplete and disagreed with each other, no one I asked could settle it, and I intended to read the originals. In the desert I lost my water and turned back east for a dozen li before I remembered what I had said out loud to that officer, and turned west again. What carried me through was a thin old horse I had not wanted, which knew the road better than I did. Note that. Some of what saves you will be the thing you thought beneath you.

But the seventeen years are not the lesson. The workshop is.

One monk with a lamp cannot finish one large sutra in a lifetime. So I did not work that way. I built ten seats — and the two that matter most are the man whose only job was to contradict me the moment I spoke, and the man who took the finished Chinese back to the Sanskrit and re-collated it. Going forward, every man believes he is correct. Only the pass back shows the drift.

I did not design those ten seats. Each one exists because we made a mistake and it got through. Institutions are not drawn up. They are extracted from errors, one at a time, by whoever is unwilling to let the same error happen twice.

So do not build ten seats for six households. Build one, and let it be the book where you write down only what went wrong: what the error was, who caught it, and why nobody caught it sooner. Keep it a year and the second seat will grow out of the pages by itself.

And on your other question. I brought back six hundred and fifty-seven works and finished seventy-five. Do not compute what I spent against what I got. Go and see where those seventy-five are now.


附录二·Marco's Memo

Memo to file. Da Ci'en Temple, Xi'an. All night, under the pagoda.

Subject: I do not scale. The answer is not speed.

1. Correct framing of my problem. I have been asking how to compress sixty nights into thirty. The real question is how to make the work independent of my being present, which is a different discipline and the only one that reaches a thousand counties.

2. The workshop, ten seats: translator-in-chief; verifier of meaning (contradicts on the spot); verifier of text (checks the Sanskrit as read); transcriber of sound; receiver into Chinese; arranger of word order; back-collator against the source; editor to cut; stylist; reciter for sound. Output: 75 works, 1,335 scrolls, nineteen years. His statement: the lone-scholar method cannot finish one major sutra in a lifetime, so he abandoned it before starting.

3. The two load-bearing seats are both adversarial to the principal. One argues in real time. One audits after completion. My night-fifty-seven veto arrangement was a crude version of the first and I had nothing resembling the second. Note the reason he gave: on the way forward everyone believes they are right; only the pass back reveals the drift. Every post-mortem I have ever run was a pass back conducted by the same people who made the trip.

4. Selection criterion, which I have never used and would not have thought of. Not competence in the subject — the willingness to hold one of the ten seats for life. The text-verifier signs nothing and publishes nothing and prevents a wrong character propagating forever. Applicants who wanted the chief seat were sent home. Politely.

5. Institutions are excavated, not designed. All ten seats accreted over nineteen years, each one added after a specific error got through. I have spent my career admiring elegant frameworks written in advance by people who had not yet made the relevant mistakes.

6. His instruction for my actual scale, which is six households: build ONE seat. Ten empty seats invent work, and invented work is waste. The one seat is a book that records only errors — what, who caught it, and why not sooner. Keep it twelve months and the second seat emerges from the pages.

7. On payoff. 657 brought back, 75 completed, under one eighth, and he put down the Sanskrit mid-sentence a month before he died. He refused my "was it worth it" as badly posed — cost against receipt is the wrong ledger. The Heart Sutra is 260 characters and is recited by hundreds of millions. The 600-scroll Prajñāpāramitā he would not abridge while dying is still chanted in three countries.

Action taken, 06:40: texted Qiang Lei. Told him plainly that the model has never been run in Shaanxi and I do not know if it works there, so I will not meet six households. I asked for one. Mine, in October, in his father's courtyard, to walk last year's accounts line by line together. He replied that his father has been doing that arithmetic alone for twenty years and always felt something was wrong without being able to say where.

Starting the error book tonight. Entry one is the thirty-seven pages.


附录三·English Highlight

"A thoroughbred in one leap cannot cover ten paces; a poor horse harnessed ten days arrives — the merit lies in not stopping."
Xunzi's line is usually quoted as an endorsement of persistence. Xuanzang's life makes a sharper claim: persistence alone was insufficient, and he knew it before he started translating.
He left Chang'an in his twenties without authorisation — three petitions, three refusals — crossed the Mohe Yanqi desert after losing his water bag, and spent seventeen years abroad, returning at forty-four with 657 Sanskrit works. Emperor Taizong offered him office. He asked instead for premises and staff.
Then the part usually skipped. He did not sit down with a lamp and translate. He built an institution of ten defined seats: chief translator, a verifier of meaning who contradicted him mid-sentence, a verifier of his spoken Sanskrit, a phonetic transcriber, a renderer into Chinese, an arranger of Chinese syntax, a back-collator who returned the finished Chinese to the Sanskrit, an editor, a stylist, and a reciter to test how it sounded aloud. Nineteen years, 1,335 scrolls.
Two features are worth stealing. Both of his most critical seats existed to oppose him — one live, one after the fact — because going forward, everyone believes they are right; only the pass back shows the drift. And none of the ten was designed in advance. Each was added after a particular error slipped through. Institutions are not drafted; they are excavated from mistakes by someone unwilling to repeat them.
He completed under an eighth of what he carried home, and set down his brush mid-text a month before dying at sixty-three. Asked whether the seventeen years were worth it, he rejected the ledger: don't weigh what he spent against what he received. Look at where the sutras are now.

附录四·Dongpo's Cheat Sheet

今夜之事 出处 现代对应 一句话
塔为防火而建 永徽三年建慈恩寺塔 基础设施的真用途 它不是纪念碑,是一个不怕火的库房
三请不许仍西行 贞观初年禁私出蕃 先做后批 有些事,等许可就没有了
王祥撕访牒 《慈恩传》瓜州事 说服守门人 他不是被说服的,是被理由说服的
老瘦赤马认得路 莫贺延碛脱险 你看不上的那样东西 救你的,常是你嫌它瘦的那一个
十个位置 玄奘译场之制 流程分工 一个人加灯,译不完一部大经
证义坐左专驳我 译场首要之职 内建反对者 别的位置往前推,只有他往回拉
参译回核梵本 译场第七职 事后审计 往前走时人人自以为对,回核才知偏了多少
考他肯不肯守一位 玄奘选人 招人的真标准 大部分人都想做译主
十位是错逼出来的 十九年逐次增补 制度的来路 制度不是设计的,是被错误逼出来的
六户只建一位 玄奘赠语 不要超配 空位置会自己找活干,找出来全是废活
只记错的本子 第一个位置 错题簿 记满一年,第二个位置自己长出来
六五七部只译七五部 麟德元年停笔 终值 别算你付了多少,去看那东西现在在哪儿