5D Library › 胰岛素胰高血糖素双稳态、内脏脂解动力学与线性规划配餐 · English

第七章 古典到底在警告什么

一处值得弄对的字句之争

每当一场中文的食疗论证需要一个古代的凭据,人们引的那句话出自《素问》第三篇《生气通天论》。在明代翻刻宋本里,字形作「高梁之变,足生大丁」,而通行的现代影印本沿此一系 (黄帝内经素问 1963)。关于它,有两件事经常被说得比文本所允许的更有把握。

原典

传本与注。 宋本一系的文字把前两个字写作「高」与「梁」。王冰唐代的注把它们释为「膏」与「粱」,所以「肥甘厚味与精细谷物」这一读法至少可以上溯到唐代;许多后出的版本直接在正文里印那两个较丰的字形。第三个字「足」,多数注家读作「足以」,于是全句为「膏粱之变足以生出大丁」;少数注家按解剖读,释为足部。传到现代食疗学里的那个意思,也就是富人的病来自他们所吃之物的丰厚,依据的是王冰的注,而不是光秃秃的字形 (黄帝内经素问 1963)。

要小心的理由在于,通常架在这句话上的论断,并不是这句话所作的论断。它被引来谴责碳水化合物。它谴责的是「膏粱」,富贵人家餐桌上的肥与精:密度与精细合在一起,而不是淀粉本身。七世纪写作的孙思邈则明确指出,食疗的主题是适宜而不是戒绝。他《备急千金要方》的食治篇开篇便把食放在药之前,并说不知食宜的人不足以存生 (孙思邈 1982)。

原典

孙思邈论食在先。 《千金要方》食治篇开篇陈述:安身之本必资于食,救疾之速必凭于药,而不知食宜者不足以存生。该篇随后逐一按食材铺陈,列其性味与宜忌,这是一种拣选的食疗学,而不是一种禁绝的食疗学 (孙思邈 1982)。

这个警告是否关于密度,可以用算术核查,因为密度是一个数。

推导

# Chapter 7. p13: What the classical warning about rich food amounts to
# in numbers. Four illustrative daily patterns are costed in energy
# density: the plain millet-and-vegetable diet of a Tang household, the
# rich meat-and-refined-grain diet the classics warn the wealthy about,
# a modern mixed diet, and a modern diet built on packaged food. For
# each, the mass of food a day of 1800 kcal takes, the chewing time at
# a fixed rate, and the carbohydrate load are reported. Composition is
# illustrative and rounded; the point is the ratio, not the decimal.
DIETS = (
    # name, kcal per 100 g, carbohydrate g per 100 g, chews per 100 g
    ("plain Tang household",   96, 17.0, 420),
    ("rich meat and grain",   233, 24.0, 190),
    ("modern mixed",          170, 20.0, 240),
    ("modern packaged",       348, 41.0,  85),
)
KCAL_DAY = 1800.0
CHEW_RATE = 72.0            # chews a minute

if __name__ == "__main__":
    print("   a day of {:.0f} kcal from each pattern".format(KCAL_DAY))
    print("   pattern                 grams   carb g   chewing minutes")
    rows = []
    for name, dens, carb, chews in DIETS:
        grams = KCAL_DAY / dens * 100.0
        carb_g = grams * carb / 100.0
        minutes = grams * chews / 100.0 / CHEW_RATE
        rows.append((name, grams, carb_g, minutes, dens))
        print("   {:<24}{:<8.0f}{:<9.1f}{:.1f}"
              .format(name, grams, carb_g, minutes))
    plain = rows[0]
    rich = rows[1]
    packaged = rows[3]
    print("   energy density, packaged against plain: {:.2f} times"
          .format(packaged[4] / plain[4]))
    print("   mass of food, plain against packaged: {:.2f} times"
          .format(plain[1] / packaged[1]))
    print("   chewing, plain against packaged: {:.2f} times"
          .format(plain[3] / packaged[3]))
    print("   carbohydrate, packaged against plain: {:.2f} times"
          .format(packaged[2] / plain[2]))
    print("   rich against plain: density {:.2f}, mass {:.2f},"
          " chewing {:.2f}".format(rich[4] / plain[4], rich[1] / plain[1],
                                   rich[3] / plain[3]))
    print("   grams of food an hour of chewing buys:")
    for name, grams, carb_g, minutes, dens in rows:
        print("   {:<24}{:.0f}".format(name, grams / minutes * 60.0))
    print("   kcal an hour of chewing buys:")
    for name, grams, carb_g, minutes, dens in rows:
        print("   {:<24}{:.0f}".format(name, KCAL_DAY / minutes * 60.0))
    print("   the classical warning is about density, and the ratio")
    print("   between the extremes of this table is {:.2f}"
          .format(packaged[4] / plain[4]))
#    a day of 1800 kcal from each pattern
#    pattern                 grams   carb g   chewing minutes
#    plain Tang household    1875    318.8    109.4
#    rich meat and grain     773     185.4    20.4
#    modern mixed            1059    211.8    35.3
#    modern packaged         517     212.1    6.1
#    energy density, packaged against plain: 3.62 times
#    mass of food, plain against packaged: 3.62 times
#    chewing, plain against packaged: 17.91 times
#    carbohydrate, packaged against plain: 0.67 times
#    rich against plain: density 2.43, mass 0.41, chewing 0.19
#    grams of food an hour of chewing buys:
#    plain Tang household    1029
#    rich meat and grain     2274
#    modern mixed            1800
#    modern packaged         5082
#    kcal an hour of chewing buys:
#    plain Tang household    987
#    rich meat and grain     5298
#    modern mixed            3060
#    modern packaged         17687
#    the classical warning is about density, and the ratio
#    between the extremes of this table is 3.62

按唐代寻常人家的模式吃出 1800 kcal 的一天,重 1875 克,带 318.8 克碳水化合物;同样的能量由厚味的肉与精粮模式供给,重 773 克,带 185.4 克;由现代的包装食品模式供给,重 517 克,带 212.1 克。所以那个古代的寻常一天,也就是文本并没有警告的那一天,反而是碳水化合物最多的一天,而且差得很远。把这些模式分开的那个量不是淀粉,而是能量密度,两个极端之间相差 3.62 倍,以及进食的费力程度,按咀嚼时间相差 17.91 倍。一小时的咀嚼在寻常饮食上买到 987 kcal,在包装食品上买到 17687 kcal。

这就是那条古典警告用现代单位说出来的样子。「膏粱」是一个密度论断。一种以《素问》之名谴责稻米与小米的食疗学,有文本站在它的对面;一种谴责密度的食疗学,有文本站在它这一边,而第三章与第六章已经显示,密度也正是让一个热量缺口好守或难守的那件事。

任何建议都躲不开的那笔账

本书每一个论断底下都有一条守恒律,而它把一个问题彻底定下来:减掉的脂肪去了哪里。Meerman 与 Brown 就以这种直截了当的形式提出这个问题并算通了化学计量,顺带指出临床医生给出的答案有多经常是错的 (Meerman and Brown 2014)。

推导

# Chapter 7. p14: The thermodynamic accounting. Fat is not released into
# the lavatory or the sweat: it is oxidised, and almost all of its mass
# leaves through the lungs as carbon dioxide. Palmitic acid stands for
# stored fat and glucose for stored carbohydrate. The balanced equations
# are C16H32O2 + 23 O2 giving 16 CO2 and 16 H2O, and C6H12O6 + 6 O2
# giving 6 CO2 and 6 H2O. From the molar masses below everything else
# follows: oxygen needed, carbon dioxide and water produced, the
# respiratory quotient, and the share of the lost mass that is exhaled.
M_PALM, M_GLU = 256.42, 180.16
M_O2, M_CO2, M_H2O = 32.00, 44.01, 18.02
ATP_PALM, ATP_GLU = 106.0, 32.0
KCAL_PALM, KCAL_GLU = 9.0, 4.0        # kcal a gram as burnt in the body
ADIPOSE_LIPID = 0.87                  # lipid share of adipose tissue mass
DEFICIT = 500.0

def per_gram(mass, n_o2, n_co2, n_h2o, atp):
    return (n_o2 * M_O2 / mass, n_co2 * M_CO2 / mass,
            n_h2o * M_H2O / mass, atp / mass)

if __name__ == "__main__":
    fat = per_gram(M_PALM, 23, 16, 16, ATP_PALM)
    glu = per_gram(M_GLU, 6, 6, 6, ATP_GLU)
    print("   per gram burnt            fat        carbohydrate")
    print("   oxygen needed, g          {:<11.4f}{:.4f}".format(fat[0],
                                                                glu[0]))
    print("   carbon dioxide out, g     {:<11.4f}{:.4f}".format(fat[1],
                                                                glu[1]))
    print("   water out, g              {:<11.4f}{:.4f}".format(fat[2],
                                                                glu[2]))
    print("   moles of ATP              {:<11.4f}{:.4f}".format(fat[3],
                                                                glu[3]))
    print("   respiratory quotient      {:<11.4f}{:.4f}"
          .format(16.0 / 23.0, 1.0))
    print("   moles of ATP per kcal     {:<11.4f}{:.4f}"
          .format(fat[3] / KCAL_PALM, glu[3] / KCAL_GLU))
    print("   fat gives {:.2f} times the ATP of carbohydrate per gram"
          .format(fat[3] / glu[3]))
    print("   and {:.2f} times per kcal"
          .format((fat[3] / KCAL_PALM) / (glu[3] / KCAL_GLU)))
    exhaled = fat[1] / (fat[1] + fat[2])
    print("   of the mass leaving a burnt gram of fat, {:.3f} is exhaled"
          .format(exhaled))
    print("   and {:.3f} leaves as water".format(1.0 - exhaled))
    kg = 1000.0 * ADIPOSE_LIPID
    print("   one kilogram of adipose tissue holds {:.0f} g of lipid"
          .format(kg))
    print("   worth {:.0f} kcal at {:.1f} kcal a gram"
          .format(kg * KCAL_PALM, KCAL_PALM))
    print("   at a deficit of {:.0f} kcal a day that is {:.1f} days"
          .format(DEFICIT, kg * KCAL_PALM / DEFICIT))
    print("   burning it needs {:.1f} kg of oxygen and breathes out"
          .format(kg * fat[0] / 1000.0))
    print("   {:.1f} kg of carbon dioxide and {:.1f} kg of water"
          .format(kg * fat[1] / 1000.0, kg * fat[2] / 1000.0))
    litres = kg * fat[0] / M_O2 * 22.414
    print("   that oxygen is {:.0f} litres at standard conditions"
          .format(litres))
    print("   or {:.1f} litres a day across the {:.1f} days"
          .format(litres / (kg * KCAL_PALM / DEFICIT),
                  kg * KCAL_PALM / DEFICIT))
#    per gram burnt            fat        carbohydrate
#    oxygen needed, g          2.8703     1.0657
#    carbon dioxide out, g     2.7461     1.4657
#    water out, g              1.1244     0.6001
#    moles of ATP              0.4134     0.1776
#    respiratory quotient      0.6957     1.0000
#    moles of ATP per kcal     0.0459     0.0444
#    fat gives 2.33 times the ATP of carbohydrate per gram
#    and 1.03 times per kcal
#    of the mass leaving a burnt gram of fat, 0.709 is exhaled
#    and 0.291 leaves as water
#    one kilogram of adipose tissue holds 870 g of lipid
#    worth 7830 kcal at 9.0 kcal a gram
#    at a deficit of 500 kcal a day that is 15.7 days
#    burning it needs 2.5 kg of oxygen and breathes out
#    2.4 kg of carbon dioxide and 1.0 kg of water
#    that oxygen is 1749 litres at standard conditions
#    or 111.7 litres a day across the 15.7 days

烧掉一克脂肪需要 2.8703 克氧,产出 2.7461 克二氧化碳与 1.1244 克水;一克碳水化合物对应的是 1.0657、1.4657 与 0.6001。呼吸商分别是 0.6957 与 1.0000,间接测热法就是靠这个把两种燃料分开的。烧掉一克脂肪所离开的质量里,0.709 被呼出,0.291 以水离开。脂肪是被呼出去的。

有两处对照值得记住。按每克算,脂肪给出的 ATP 是碳水化合物的 2.33 倍,但按每千卡算只有 1.03 倍,这正是「一千卡就是一千卡」成立的那个意义上:能量密度不同,而每单位能量的效率几乎相同。还有,一公斤脂肪组织含 870 克脂质,值 7830 kcal,在每天 500 kcal 的缺口下需要 15.7 天,并需要 1749 升氧,也就是比身体本来会用的每天多约 111.7 升。

假说 · 尚未验证

作者的诠释(尚未验证): 古典范畴「膏粱」与现代的能量密度加进食省力程度的对应,比它与任何宏量营养素范畴的对应更紧;一个密度与费力的合成指标,会比碳水化合物含量更好地排出传统饮食禁忌的次序。什么会证明它错: 如果对《素问》、《千金要方》及后世食治文献中的饮食禁忌作系统编码后,所提到食物的碳水化合物含量比其能量密度与咀嚼费力程度更能预测禁忌,那么这一读法就是错的。

小结

《素问》宋本一系写的是较素的字形,而王冰唐注提供了「膏粱」这一读法,所以这个警告关于密度与精细,而多数注家把第三个字读作「足以」而不是足部。算术也同意:古代寻常的一天带 318.8 克碳水化合物,比厚味那一天还多,而两极之间的能量密度相差 3.62 倍、咀嚼相差 17.91 倍。热力学账目显示烧掉的脂肪有 0.709 从肺里走。第八章把这个月拼起来,然后算出谁不可以尝试它。

参考文献

Meerman, Ruben, and Andrew J. Brown. 2014. “When Somebody Loses Weight, Where Does the Fat Go?” The BMJ 349: g7257.
孙思邈. 1982. 备急千金要方. 人民卫生出版社.
黄帝内经素问. 1963. 人民卫生出版社.

Fan, Y. (2026). Insulin-Glucagon Bistability, Visceral Lipolysis Kinetics and Linear-Programming Diet Optimization: Hepatic Glycogen Depletion Times, Triglyceride-Glucose Indices and Micronutrient-Constrained Menus [胰岛素胰高血糖素双稳态、内脏脂解动力学与线性规划配餐] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090248 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).