戴维斯
编者按
冷战期间,核武器试验在进行的同时,国际上对于是否应该禁止核试验的讨论也在继续。禁止核试验的障碍之一是如何证实一个国家是否遵守此禁令。苏联特别不情愿外界对其核武器设施进行监测,因此,证实是否进行核试验将需要依赖于从远处监测核试验的能力。地下核试验直到20世纪70年代早期还在进行,试验会产生地震波,但是利用这些地震波进行核实的话需要将其和天然地震区分开来。本文中,麻省理工学院的戴维·戴维斯综述了地震学识别的进展,并得出结论,该技术已经能够胜任监测需要,但还不是绝对可靠的。 英文
在过去的十年里,监测和识别地下核爆炸的地震学方法取得了稳步发展。这门技术可以将爆炸和地震区分开来。目前的问题是要降低阈值和理解偶然性的有问题事件。 英文
一次地下核爆炸所释放的能量中,有百分之一转化为地震波,而这些地震波中就携带着关于这一事件的时间、规模和地点等信息。它们也可以说明,这一事件确实是一次爆炸而不是地震。这种信息的提取还要更难一些,因而在过去的几年中,很多国家的研究都集中指向了对天然事件与人为事件的识别。还不知道有其他识别地下爆炸的方法能够应用于全世界范围。在本文中,我将描述这一学科在最近所取得的进展,这些进展对于武器控制具有如此明显的作用。但是,我所得到的结论并不能被解释为完全禁止核试验前景的一种度量。促进条约制定的因素有很多,监管能力只是其中之一。由此看来,地震学方法的作用并不足够强,但是,至少具体到某些爆炸当量水平时,它可能是必要的。如何制定出合适的水平标准目前还是一件有争议的事情。这里所报道的是很多地震学家的研究成果。不过,为了将参考文献数量减少到可控制的限度,我只引用了一些大型的研究结果汇编,其中汇集了各种研究思想和研究工具。 英文
在地下进行核武器试验的原因,显然不是一个公开讨论的问题,不过尼尔德和鲁伊纳的论文中基本上为此提供了合理的完整解释 [1] 。在1968年至1971年间,美国宣称平均每年进行大约25次地下核试验,与之相比,苏联每年大约进行10次核试验,中国每年1次,而法国大约每年5次——不过中国所进行的除1次以外的所有试验和法国进行的所有试验为地上试验。英国已有7年没有宣布过核试验。 英文
1958年,禁止核试验首次成为国际问题,在日内瓦举行的关于技术问题的会议明确指出,地震学需要取得实质性的进展,使得仪器观测毫不含糊地指出具有几千吨(kton)当量水平的爆炸的发生。确实,那次会议和随后的商谈中可用数据甚少(当时只有两次或者三次规模很小的地下试验),因此,首要目标是建立包括大约170个台站的国际台网,该台网能够检测到一定阈值之上的地震信号,但是不必确定其来源。人们期望,绝大多数地震事件可以通过其位置、深度、辐射模式和信号形状明确地体现出天然地震的特征,但还剩余少数事件有待于进一步研究。在早期的国际磋商过程中,详细讨论了如何对一小部分可疑事件发生的地点进行检查——检查的数量和性质是争议较大的问题。同时,苏联对于检查问题的立场渐趋强硬,并最终断言检查是不必要的,而单纯的国家监管方式就能满足需要了。因此“日内瓦国际台网”从未曾建立起来。 英文
1958年时所关注的两个中心议题到今天仍然是焦点。地震背景噪声(来自风、交通工具、海浪等)限制了地震事件检测的进行;而且某些地震也不是马上就显现出天然地震的性质。自1958年以来,地震学已经取得了显著进展,不过提高信噪比(s.n.r.)和寻找区分地震与爆炸事件的改进方法仍然是主要研究领域。地震学中所涉及的极为狭窄的频率段严格地限制着这项技术,由于研究工作集中于低信噪比条件下,因此关键是找到具有最高信噪比的频率,集中在该频率下取得进展,而不是试图包揽震动信号的全部图谱。地震学中这种“任务导向式”的方法,再加上研究过程中产生的理性所激发出的问题,对于地震学自1958年以来取得的发展具有重要的作用。 英文
在日内瓦研讨会议的早期(该会议持续到1962年)就已明确,无论国际约定最终将如何发展,国家利益需要的是提高地震学能力。尤其是美国、英国和瑞典,出于明确的禁止核试验的监控目的,开始提高其地震学能力以适应现代化需求。其他国家,尤其是加拿大和日本,也在提高其地震学能力。最为浩大的工程无疑是美国的“船帆座计划”,目前已经为其研究和发展花费了超过1.5亿美元。收获的是什么呢? 英文
在最初的几年中,基础研究得到大力支持,由100个地震台站组成的台网,即世界标准地震台网(WWSSN),建立起来——不过,没有一个地震台站是位于苏联或其盟友的疆域中的。这个台网提供了庞大的数据库,统一地记录了短周期(约1 s)和长周期(约20 s)的地震信号。 英文
一些“事实”迅速从研究工作中显现出来。对于苏联加入国际台网的可能性逐渐降低使得远震研究居于主导地位;在距离震源只有几度的范围内可以很好地记录地震体波(P波和S波);此后到大约25°的范围内,体波信号质量一般且变化大;此后在远达100°范围内体波都是清楚和可判定的。距离以球心角度数来测量。图1中给出了一些P波的实例。最终,远震区域可以覆盖半个世界,这在某些国家不允许进入时显得至关重要。观测区域是受限制的,这直接限制了对爆炸的检测能力。一次发生在硬质岩石(例如花岗岩或盐岩)中的5 kton当量水平的爆炸,在远震区域中产生一个P波,其振幅大约为5 nm,周期为1 s。在最佳观测中,1 s周期的背景噪声约为该信号的三分之一。因此,在没有噪声剔除技术时,监测台网一般可以检测到硬质岩石中发生的5 kton当量水平的爆炸(直到有四个彼此间隔较大的地震台站记录到信噪比至少为1.5的信号才能说一次震动事件被检测到了)。低于此当量水平时,检测能力剧烈下降。此外,发生在较软岩石(尤其是凝灰岩,在内华达很常见的一种火山灰)中的爆炸产生的信号较弱,与硬质岩石中发生的同级别爆炸相比,信号相差至少两倍。发生于干燥冲积层中的爆炸产生更小的震动效应——硬质岩石中的1 kton爆炸与冲积层中的10 kton爆炸基本相当。不过,在地球上任何地方发现的冲积层都有厚度的限制,对于一次基于安全考虑而必须充分深埋的爆炸来说,在已知的冲积层中不大可能引爆和容纳一次超过20 kton当量水平的爆炸。 英文
图1. 短周期远震的P波观测结果,震动事件位置为:a,苏联东哈萨克;b,阿拉斯加;c,土耳其;d,中国新疆南部。顶部和底部的事件被推测为地下爆炸。所有上述事件的体波震级介于5.0至5.7之间。注意,这里最大的困难是识别事件中的初动方向。
在WWSSN配备有常规仪器和照相记录装置的同时,还进行了若干台阵实验。其目的在于通过增加多通道来抑制假定不相关的噪声。另外一个优点是将数据记录在模拟磁带和后来的数字磁带上。尽管实际上地震学中的所有决定都是基于肉眼观测而做出的,但是直接的直观记录还是在动态范围内遭受到明显的不利影响。此外,在大约6 s周期出现一个源于海洋的微震所产生的强噪声峰值,而峰值周期在1 s和20 s的WWSSN仪器装置对于该噪声的限波,通常也不能完全让人满意。结果是,通过观察其放大率可调控的胶片记录,6 s微震常常占据主导,而1 s和20 s信号则因此无法达到最佳放大。对磁带记录进行数据处理的可能则消除了这一障碍。 英文
建于美国的第一批台阵的孔径为4 km。它在很多方面体现了日内瓦会议对于国际台网阵列的设想。在此孔径范围内布设了多达16台的地震仪,来自所有地震仪的信号不经过相位调整就叠加起来。远震信号到达时信号明显变化——信号以至少10 km·s
–1
直至24 km·s
–1
水平速度沿地表传播,在典型的场地对应的入射角小于20°。因此,直接叠加不会严重减弱小孔径台阵观测到的1 s信号。通过这些小型台阵发现,特别是当仪器间距大于0.5 km时噪声在这一频率段是相干的,因此信噪比的增益将介于1(如果噪声是完全相干的并且沿近垂直方向传播)和
(如果噪声是完全不相干的)之间。后来的台阵使用的地震仪间距大约为2 km,在这一距离,本地噪声是显著不相干的。
英文
下一步发展是由英国原子能管理局装配的中等孔径的相控阵列
[2]
。多达20台地震仪放置于两条相互垂直的直线上。这种类型的台阵还安装在加拿大、巴西、苏格兰、印度和澳大利亚。孔径大到需要在记录信号后对其进行相位调制,以将台阵操控到合适的方向上,但又小到足以保证信号仍然是相干的。由此获得了接近于
(约为4.5)的信噪比增益。
英文
第一个大孔径地震台阵(蒙大拿的LASA)是由美国于1965年完成的,第二个台阵(挪威的NORSAR)也已工作了两年左右。LASA的孔径达到200 km,包含350个短周期地震仪;NORSAR则要小一点。LASA定向(由一台计算机来实现)地震信号的入射方向,并很好地确定入射角度,从而能够以100 km或200 km的精度定位信号源。LASA和NORSAR的实时数字操作要各自占用整整两台中型计算机来搜寻P波。不过,大型台阵的增益不能达到
的预期,因为孔径超过200 km的地壳地质条件会出现很大的变化,这将导致信号相干性的减弱。不过,对于LASA而言信噪比增强10倍,并且该台阵能够在远震范围内对硬质岩石中至少2 kton当量水平的爆炸进行检测和定位。P波振幅与此相当或大于此水平的地震,每年会发生几千次,它们构成了必须与爆炸相区分开的事件群。
英文
到20世纪60年代中期,体波记录不足以解决识别的问题越来越明显;必须要有频谱更宽的地震资料。答案就是长周期地震道的优势信号,瑞利波。这是一种沿着地表传播的弥散波,在20 s周期处被充分激发——有短周期的瑞利波和有长周期的P波,不过它们对于识别问题无关紧要。对于短周期记录来说,噪声是限制因素。6 s的海洋微震背景,由于大气效应所产生的长周期波动,以及其他地震所产生的瑞利波列的干涉,都会对背景有贡献。在没有干涉事件的情况下,大多数地震台站在20 s周期的背景噪声水平为50 nm到100 nm;噪声水平会随着地点和时间的不同而大幅度变动。由已给出的理由来看,磁带记录是最合适的。 英文
瑞利波呈圆柱状传播,但在大地中也发生其他形式衰减。信号随距离(Δ)的衰减指数为1.6。这种快速衰减就要求地震台站必须尽可能的接近事件发生地。在非侵犯性监控的政治要求范围内,我们只得考虑以20°的Δ值作为可用到的地震台站的最小距离。在20°时,周期为20 s的10 kton爆炸所产生的瑞利波振幅略小于100 nm,因此在此距离单个地震台站检测到面波的可能性是很小的。就短周期观测而言,台阵证实为一种提高信噪比的有价值的方法。仪器间距必须较大(典型距离为20 km)才能使噪声不相干。在瑞典、加拿大和印度,小的三重台阵有的正在兴建中,有的已在最近建成。LASA、NORSAR和ALPA(位于阿拉斯加的专用的长周期的台阵)都至少有15个长周期地震仪。LASA得到的结果表明,信噪比增益系数增加了4倍——接近于
。此外,对于弥散波性质的了解使匹配滤波器技术得以应用,而这些使信噪比又增加了2倍。因此,一个由二十个地震仪组成的长周期台阵能使信噪比相对于不相干噪声的增强达一个数量级。除此之外,台阵还有一定的方向识别能力(瑞利波在20 s周期具有80 km的波长),这就使得研究干涉信号之中的事件成为可能,条件是两个信号来自于相差30°以上的方向,并且干涉信号强度不超过所研究信号的十倍。
英文
仪器制造领域最近取得了新的进展,如通过更为有效的密封和隔离技术来减少仪器中的非地震噪声,以及扩展仪器响应范围以覆盖广泛的宽频谱。另一种有前景的做法是将仪器置于往往可以明显降低长周期噪声的钻孔之中。 英文
P波和瑞利波都用于定义地震事件的震级。基于里氏标度的体波震级( m b )是对P波激发程度的量度。将记录中的P波信号振幅转化为卓越周期 T 时的地面位移( A ),接着应用于下面的公式:
其中, B (Δ)项用来补偿震中距离不同而导致的信号变化。例如,震中距离60°处周期1 s内的地面位移为10 nm意味着震级为4.7的事件。 英文
面波震级是通过测量周期为 T (通常约为20 s)的瑞利波的地面位移 A ,并应用下面的公式而得到的:
M s 震级4.0的事件,在30°距离和20 s周期处将有300 nm的地面位移。 英文
对于地震来说, M s 与 m b 之间有经验性的、离散较大的关系,其最佳拟合式为:
M s =1.59 m b –3.9
后面我将进一步讨论这一关系。 英文
将 M s 和 m b 与爆炸当量联系起来是可能的。硬质岩石中的2 kton爆炸产生 m b 为4.0的地震事件,20 kton爆炸则产生 m b 为5.0的地震事件。爆炸当量提高一个数量级对应于震级加一的关系因为几个理由而无法成立。我曾经提到过较软的岩石产生较小的地震信号; m b 为4.0的事件相当于花岗岩中的2 kton爆炸,凝灰岩中的3 kton或4 kton爆炸以及冲积层中的20 kton爆炸。 英文
同样的,面波震级 M s 也可以与爆炸当量较好地联系起来,而且还是更为有用的一种量度,因为看起来至少在1 kton到1 mton范围内是遵循如下形式关系(对于硬质岩石而言)的:
M s = 1.3 log 10 Y + 1.5
在这一研究课题的早期发展历程中,对于通过P波检测事件的关注引起了一种希冀,即只凭对P波的研究就可以揭示出爆炸与地震之间的区别。人们很快发现这种区别确实是存在的,但还没到足以把所有的爆炸与所有的地震分离出来的程度;它只能提供可能的概率。我们所说的辅助性判断,是事件发生的方位和深度,地震道的初动方向,以及信号的复杂性。很明显,如果一个事件明确地确定为发生在几千米以下的位置,它就不是一次爆炸——如果发生于深水之下也是同理。深度测定的精度随着深度下降而降低,通常在低于30 km的地方就无法确定事件发生的深度了。初动准则利用爆炸与地震辐射图案的不同,前者辐射图案是完全压缩式的,后者则是压缩与膨胀相继出现在相邻象限中。有许多问题——地震台站在全球的分布不均以及极性测量所需要的高信噪比,这些都会使该技术对于低当量水平事件无计可施。“复杂性”是对P波信号持续性的量度,是一项有前景的判据;持续时间超过几秒钟的信号可能来自于地震。遗憾的是,有很多地震持续时间不到一秒钟,它们只能被归入“可疑事件”的类别中。 英文
对P波频谱更新的研究指出,即使是震级很小的爆炸,其高频段(2 Hz到3 Hz)信号也比地震要丰富很多。这与人们普遍接受的地震和爆炸震源特征的观念是符合的。但是,区域性变化带来了问题;目前看来,上地幔衰减(也就是频率传输特性)随着地区不同而显著不同,地震和爆炸频谱的真实差异被上地幔的频率吸收所掩盖,以至于爆炸的P波有时候看起来像地震。 英文
瑞利波与P波的联合研究产生的频谱拓展为该研究主题带来了一项毋庸置疑的“突破”。人们在几年前就已认识到,与地震相比,有相同体波震级的爆炸产生不了同样震级的瑞利波。WWSSN使得这些观测有了一个更定量化的和全球化的基础。可以看到,如果对每一事件以表面波震级 M s 对应于体波震级 m b 作图,天然和人为事件将分别汇聚在两条明显分离的直线周围。 英文
图2中给出了一个典型的实例。很明显,地震事件总体比爆炸事件总体更分散一些,考虑到地震类型的多样性,这并不令人惊奇。实际地球中复杂传播过程的不确定性也导致了这种分散。但是有一点可以确定——这幅图大大增加了我们识别超过一定规模的事件的信心。当然,纯粹主义者有理由反驳,类似图2这样的数据图并不能提供确定性,因为它只包含大小适中的样本集合;尽管如此,这幅图还是得到了广泛认可。 英文
图2. 马歇尔和巴沙姆基于WWSSN数据的亚欧大陆事件的Ms : mb图。○代表地震,●代表推测出的爆炸。
地震和爆炸还产生其他地震相——剪切波或称S波,以及被称为勒夫波的水平极化表面波。此外P波中还有一长周期成分。所有这些都为识别不同的事件提供了可能性,特别是由于在理论上爆炸是几乎不产生S波和勒夫波的。对于大规模爆炸和地震而言,利用这些波的效果很好,不过人们发现,一般来说,长周期记录中振幅最大的是瑞利波。 英文
图2来自于最近一次只利用WWSSN仪器的研究 [3] 。每个点代表欧亚大陆上的一次事件,每一个被列出的事件必须要被至少四个相隔较远的地震台站检测到P波。这就为所报道的事件设置了一个体波震级的下限。有几个理由能说明为什么这一界限应该是模糊的,不过作为粗略的近似,可以说,全球地震仪台网(不包括台阵)可以通过四个或者更多地震台站的P波检测到几乎所有 m b 大于等于5.0的事件,而基本上检测不到 m b 小于等于4.0的事件。 英文
每个点还代表了确定的面波震级,而且按照习惯,在接受一个数值之前需要至少有四个地震台站检测到面波(这是对地震辐射图案加以认可的一种十分粗略的方法)。 M s 为4.0的事件肯定能因其瑞利波而被检测到,但是 M s 为3.0的事件就不大可能被检测到了。这在一定程度上是由于噪声的缘故,不过接收器的位置也很重要,因为它们需要尽可能靠近事件发生地。 英文
检测阈值是指一次地震有可能同时被P波和瑞利波检测到,或者都未检测到。这意味着,几乎不会有这样的地震,即测得了 m b 但由于没有测得 M s 而无法在 M s : m b 图上加以绘制。另一方面,有可能存在这样的爆炸,比如说, m b 4.5和(通过外推法) M s 2.5的爆炸。如果需要两个参数同时确定才能识别一次事件的话,就可以讨论以识别阈值作为瑞利波的检测阈值。这一阈值在本质上是统计的,但可以通过如下方式进行定义:“如果P波检测到的事件的 M s 大于等于3.2,其面波能被四个或者更多地震台站检测到的可能性有90%,因此可以将之归为爆炸或地震中的一类。”这一稍嫌冗赘的陈述是必不可少的,因为事件的常规检测采用体波,但是关键的识别检测则用面波。另外一种描述是:“硬质岩石中发生的大于等于20 kton当量水平的爆炸同样地可以识别;因此,在实践中,WWSSN可以检测到几乎所有的地震事件。” 英文
从图2中可以明显地看出,任何要降低当量阈值的尝试,都要合理的从尝试改进面波检测着手,理由是显而易见的,俄国有一些 m b 小于5.0的事件,目前的仪器无法看到它们的面波,但是可以假定它们是爆炸。显然,随着增强检测面波能力的进程的发展,体波检测能力的提高也被寄予厚望,从而能够检测到几乎所有 m b 大于4.0的事件。后面将讨论这一策略。 英文
图2清晰地显示出对于两种类型事件,20 s瑞利波与1 sP波激发之间的差别。一个很自然的问题就是,这一差异是如何产生的,这不是纯理论性的问题,因为在努力降低检测阈值(也就是引入具有更小 M s 与 m b 值的点)的过程中,准确地知道两个样本总体在何时合并是非常有价值的。图2中是存在明显分隔的,但是在震级更小时,最小二乘拟合直线可能不是两者关系的最好表示法。关于地震和爆炸的理论模型应该可以适用于任何震级,并且可以说明,用于提高检测能力的资本投入是否是合理的。 英文
遗憾的是,我们对于两种类型事件本质的了解仍然是模糊的。众所周知,地震是破裂沿断层面传播,导致了穿过断层的有限错位;而地震仪所观测到的爆炸则相当于在半径不到1 km的假想表面上突然施加了一个弹性应力。在某种程度上我们知道P波和瑞利波是如何从两种类型的震源产生的。这还不足以回答为什么可以识别这一问题。这个问题可以通过能量和震源谱加以细分,也就是说识别可能取决于震源的能量体波和面波的不同配分方式;或者取决于震源频谱的差异,这种差异体现在与波的类型无关的1 s和20 s周期激发的差异上;或者是上述两者的某种形式结合。 英文
我认为我们对这一问题还没有找到一个简单的答案。这反映出迄今为止我们对于如何满意地描述震源发生的过程仍然无能为力。我有理由认为,掌控识别能力的既不是深度也不是尺寸的差异。除此之外就难讲了。 英文
对于那些位于全球台网覆盖之外的较小震级事件这一难题来说,一种更为直接且目前更为成功的方法就是在能够有机会接近事件位置的区域进行试验工作。因此,很多研究集中在观测内华达试验区发生的爆炸以及美国西部的地震。艾文登 [4] 报道了研究结果,即使是使用 M s 和 m b 的标准定义,在观测到的最低当量处,分离还是明显的。如果仔细地考虑区域传播效应的话,分离还能得到进一步改善。 英文
这些结果表明,在对事件区域相对无限制的接近这一条件下,美国西部的爆炸和地震能够在比硬质岩石中20 kton这一全球阈值低得多的水平上加以识别。看起来,要是没有遇到无法克服的重叠问题的话,震级改善程度可能提高一个数量级。最要紧的是,这应该理解为是在一定条件下的。例如,对苏联的监测就要放弃在其附近建地震台站的想法,并努力以高级仪器、台阵和数字化过程来补偿损失。此外,也无法保证,地震和地震波在苏联的传播与在美国西部的传播有可比性(尽管有理由持乐观态度)。但是有理由期待,降低检测台网的阈值会导致较低的识别阈值,从科学上讲,对于这一改进工作的大量投入是合理的。 英文
若干仪器的研发和安装使用目前正在进行中。为数不多的高质量长周期地震仪目前正在装配中(位于西班牙、以色列、泰国、澳大利亚、阿拉斯加和新泽西州)。另外一些适用于深埋在可以明显减弱噪声的钻孔中的仪器,目前正在生产中。位于阿拉斯加和挪威的两个新的大型长周期台阵,以及位于加拿大、瑞典和印度的三个由三个地震仪组成的长周期台阵,目前正在运行中。希望在两年内,通过来自这些仪器的大量数据评定新的识别水平。看来有理由期待,基于这些仪器,阈值至少能减少一半,即硬质岩石中10 kton当量水平。这相当于说,在监测苏联时,几乎所有 M s 大于等于2.8的事件都会在四个或更多个彼此远离的地震台显示出可检测到的面波。这是一个很适度的目标;只凭两个大型台阵加上泰国和以色列的新建地震台站就可以实现这一目标。 英文
不过,相当低的阈值也是可能达到的。对美国西部研究的外推指出,期待识别能力达到 M s 为2.0(硬质岩石中2〜3 kton)是合理的。如何才能达到这一目标呢?如果监测台网可以充分扩展,那么被监测的国家就必须想尽办法将其试验场移动到尽可能远的位置。对于苏联来说,有可能在距离监测台网中最近的地震仪30°的位置进行试验,而在这一距离, M s 2.0的爆炸所产生面波的峰峰值小于10 nm。几乎没有一个地震台站的噪声水平能够低达该峰峰值的十倍,而十倍可能是最大台阵所能提供的信噪比增益上限了。 英文
现阶段,经济对于阈值的限制可能超过了地震学。例如,最近加利福尼亚大学圣迭戈分校的布龙教授估计,将地震检测能力升级到这样高水平的整个计划将花费至少两亿美元。 英文
如何才能使一个重要计划合理发展?WWSSN目前的运作明显还不够理想,因为监测台网既不是针对识别而设计,也没有致力于识别。对某些地震仪的重新选址,特别是远离人群活动中心,将降低噪声水平。进一步降低地震仪受环境变化的影响,特别是压强和温度的波动变化,不无裨益。其目标应该是只记录到大地噪声,为了确定仪器产生的全部噪声仅限于此,是值得花费大量时间和金钱的。将一些仪器置于钻孔中也必定会进一步减少噪声。相比于照相记录,数字记录因能采用前置滤波最佳地消除6 s微震而倍受青睐。 英文
目前WWSSN生成了数据,但还没有得到分析。一个地震分析小组,既能获得高质量数据,又能得到来自大型台阵和其他被关注其贡献的地震台站的信息,将极大地增加体波检测到的事件数量,并给出体波和面波观测都最有效用的地震台站。这些知识将有助于决定下一步该做什么。 英文
可能某些地震台站还值得进一步改善。要找到这些位置,只靠测量噪声是不够的。地震学并不完全是一门可预测的学科,在某些地点,信号不知何故异乎寻常的高。这可能是由特定场地原因造成的,根据噪声水平就能排除这一理由。改进工作可以采取以好的地震台站为中心建立台阵的形式。在条件好的位置兴建新的地震台站也是有意义的。 英文
再下一步就不容易推测了。一旦在地球上重新装备仪器,人的创造力将是无限的。不过,政治和经济现实限制着监测台网的扩张。前面所勾勒出的远景可能会在五年内实现。 英文
图2的相对简单性给人造成的印象是,识别结果可以制成指示列表提供给地震记录的分析人员。在大多数情况下这是事实,但是还是有令人烦恼的问题——有些是天然地震,有些则是人为的。 英文
在考察某一区域中的事件群时,某些事件是无法基于 M s : m b 识别方法进行归类的,因为它们的面波被其他无关的地震面波所屏蔽。除非找到一种方法将两个事件分开,否则它们就不能用 M s : m b 方法进行识别。这时就非常需要一个台阵,即使如此,也只有在更早时所讨论的情况下才有成功的可能。如果屏蔽事件很大(其 m b 大于等于7.5),其在地球内部的混响甚至会使全球台网被遮盖一个小时以上,在此期间不可能检测到实际发生的任何事件的P波信号。这就为在一次大地震后立即进行爆炸的规避行为提供了可能性。目前还很难找到增强规避行为检测能力的地震学方法。 英文
另外一个自然灾害就是类似爆炸的地震。如果将图2中所用资料的时间间隔展开,最终将发现震级较大(例如 m b 为5.0)但面波几乎不能被察觉的地震。基于 M s : m b 判据,这些来自地表非常有限的几个区域的地震只能归类为爆炸。我们相信,它们可以被理解为应力降异常高的地震。随着仪器的改进,无论这些事件具有什么类似地震的特征,都会暴露出来。目前尚不清楚,随着体波检测能力的提高和较小事件数量的随之增多,是否就会发现更多的此类地震。 英文
主要的人为问题(除了故意地利用地震遮掩之外)是解耦。在一个足够大的洞中引发爆炸所辐射出的地震信号比完全夯实条件下的信号至少弱100倍。因此,如果在半径为80 m的深洞中引爆,一次50 kton爆炸就会低于检测界限,在半径小于80 m时将会发生部分解耦。显然签订协议后如此大兴土木可能会引起充分关注,目前盐矿中可能还存在可以用于解耦的洞穴。似乎还没有地震方法可以检测解耦事件,因此签署条约的成员国将不得不权衡所带来的风险。 英文
尽管有希望将检测当量下限减少到硬质岩石中的几千吨的程度,我们必须意识到违约者不太可能会在花岗岩中引爆。由于硬质岩石中2 kton的阈值将是很难达到的,也就可以说,所有在冲击层中进行的低于20 kton的试验都将是无法识别的。很幸运,这种令人沮丧的情况可能不会出现。冲积层的沉积厚度限制了爆炸深度,但另一方面,爆炸必须在足够深度引发才能避免形成沉降坑。在冲积层实验之后往往形成弹坑,也就存在明显的被检测到的危险。因此违约者将不得不考虑在冲积层引爆的风险,并且可以充分肯定,安全试验的当量水平是远小于20 kton。 英文
对于 M s : m b 识别技术感到隐约的不安是合理的。这项技术总是把大量科学技术置于一份图表之中,而且还可能被错误地应用。例如,尽管图2中所示的地震事件具有相当的代表性,其中只能包含有限的爆炸事件;其中大多数来自于一个试验场所,因此不是真正的独立样本。再说,试验场所是可以随意移动的,因此识别结果也只能代表过去而不是将来。只有当我们看到识别科学很好地建立起来时,才能减少疑虑。 英文
另一方面,还可能存在着对于 M s : m b 识别方法不够乐观而产生的一些不安,而这是值得核验的。如果一次事件同时具有可检测到的P波和瑞利波,它就能置于 M s : m b 图中。但是,如果只检测到P波却没有看到瑞利波,就不能被接受。例如,苏联境内一次 m b 为4.5的爆炸就不会产生可检测到的瑞利波。能否根据其缺乏瑞利波而称之为爆炸?很明显,这个问题取决于是否所有 m b 为4.5的地震都有可检测到的瑞利波(我们排除了事件遮蔽的问题)。除了前面提到的数量有限事件之外,基于现有的经验,答案是确实如此。对于 m b >4.5的地震有大量的 M s 数值展布,但是没有向下延伸到3.2。 英文
到目前为止,人们一直是带着怀疑的目光来看待所谓“反面证据”的。其理由很可能是,对科学家而言,一个信号的存在和缺乏同等重要,而在一个具有不同思维模式的政治家看来,“反面证据”听起来就像在犯罪现场缺少指纹一样不足以令人相信。很可能正是因为 M s : m b 图的简洁性,而过分强调了确定 M s 和 m b 的必要性。 英文
下面是关于“反面证据”类型的识别方法阐述。“当P波检测到一次事件并确定其体波震级大于等于4.5时,如果 M s ≥( m b –1.3)就将它归为地震,如果 M s <( m b –1.3)或者无法确定 M s 时就将它归为爆炸,不过前提是在噪声观测的基础上必须小于 m b –1.3。”这将使阈值转变为体波震级4.5或者硬质岩石中的5 kton,单独使用 M s : m b 识别方法得到这一结果需要耗时数年且耗费巨资。尽管这一陈述将使前面所描述的那些异常地震事件被当作是爆炸,这也同样适用于 M s : m b 技术。对于这些事件来说,一种实用方法是必要的。要建立反面证据的普适性,必须对大量样本进行详细调查。这样的工作目前正在进行中。 英文
我将以一项最近在实验室以外进行的活动作为结束。1968年,斯德哥尔摩国际和平研究所(SIPRI)召集了一次来自东西方十个国家的地震学家参与的会议,会上回顾了自1958年以来的技术进展。后来出版的SIPRI进展报告,最新一期为1971年9月(参考文献5),记录了最新的技术内容。 英文
在美国,召开了众多的技术会议。本文的讨论中大量引用了1970年伍兹霍尔会议上出现的论文(参考文献4)。最近一次会议是在马萨诸塞州的剑桥,会议认为,一些预测能力得到了强化并加深了对于问题事件的认识。 英文
联合国裁军委员会(曾几易其名)协商会议定期于日内瓦举行,期间会经常讨论地震学和禁止核试验的问题。1970年,在加拿大的倡导下,联合国对承诺提供数据的地震台站进行了登记注册,随后,一项加拿大的调查考察了这一地震台网的理论潜力 [6] 。同年,英国向日内瓦提交了一份建议,内容是关于一个能提高识别能力的新地震台网 [7] 。在国际上对于禁止核试验问题日益关注的背景下,1971年,在日内瓦召开了一次联合国技术会议。 英文
很明显,在可预知的未来,对于广大公众而言,这将会是一个一直保持活跃的主题。在这段时期内,地震学不会有什么惊人的进展,但是对于地震学的持续投入无疑将导致检出阈值的稳步下降。阈值最终将落在何处,以及是否会签署条约,将取决于多方面的因素,地震学只是其中之一。 英文
这一工作由美国国防部高级研究计划局发起。 英文
图2为巴沙姆和马歇尔友情提供,其中凝聚了他们 [3] 在加拿大渥太华能源、矿产和资源部所付出的努力。 英文
(王耀杨 翻译;吴庆举 审稿)
J. A. Huberman et al .
Editor’s Note
Nearly 20 years after the structure of DNA was published, molecular biologists were still struggling to understand how DNA, the genetic material, is replicated in living cells of higher organisms (eukaryotes). The interest of this article is that it shows how rudimentary were scientists’ ideas of that process. Huberman et al . studied HeLa cells (a familiar strain of human cells derived from a cancer) during the process of mitosis (cell division). “S” phase is that during which a cell about to divide accumulates the material eventually needed to make two cells. They conclude that DNA is first synthesized near the nuclear membrane but that it later migrates to the whole nucleus. 中文
DNA replication can occur throughout the nucleus and is not restricted to the inner surface of the nuclear membrane. 中文
THE involvement of a membrane site in DNA replication was first suggested by Jacob, Brenner and Cuzin 1 in their “replicon” model for replication of bacterial DNA, but the evidence accumulated since then is inconclusive. In prokaryotic cells, co-sedimentation of DNA replication points and cell membranes has been demonstrated 1-4 , and the origin of replication and the membrane found to be associated 5 . A lipid-free replicating DNA-protein complex from E. coli has been isolated 6 and it has been reported that only the origin, but not the growing points, of the E. coli chromosome is attached to membrane 7 . 中文
In cell fractionation experiments with mammalian cells it was also found that replication points, detected by pulse-labelling with 3 H-thymidine (TdR), are associated with the nuclear membrane (or some other large, light, hydrophobic cell structure) 8-11,13 . Association between replication points and the nuclear membrane has also been detected by electron microscope autoradiography of thin sections through pulse-labelled, unsynchronized HeLa cells. The label associated with the membrane apparently moved into the nuclear interior after a 1 h chase 13 . 中文
On the other hand, most electron microscope autoradiographic experiments suggest that replication can take place throughout the nucleus. For instance, Comings and Kakefuda 14 generally found grains located throughout the nucleus when unsynchronized human amnion cells were pulse-labelled for 5 min or more with 3 H-TdR and then sectioned and autoradiographed. When, however, cells that were supposedly synchronized at the beginning of S phase were pulse-labelled for 5 or 10 min, grains were found predominantly over the nuclear membrane. Comings and Kakefuda concluded that initiation of replication takes place on the nuclear membrane, whereas the replication occurs anywhere within the nucleus. 中文
Somewhat different results were obtained by Blondel 15 , using KB cells and pulse times of 2 min; Williams and Ockey 16 , using Chinese hamster cells and pulse times of 10 min or longer; and Erlandson and de Harven 17 , using HeLa cells and pulse times of 15 min. All three groups concluded, in agreement with Comings and Kakefuda 14 , that during a large part of the S phase grains are produced over the entire nucleus, but they differed from them in finding a peripheral pattern of grains more frequently at the end of the S phase than at the beginning. A cell fractionation experiment by Kay et al . 18 also demonstrated association of late-replicating but not early-replicating DNA with the nuclear membrane. 中文
Regardless of the time in S phase at which replication occurs close to the nuclear membrane, one important implication of most of these electron microscope autoradiography experiments is that, in some parts of S phase, at least, replication occurs throughout the nucleus. How can this implication be reconciled with the experiments indicating that all replicating DNA is associated with the nuclear membrane? It is possible that a considerable amount of DNA is synthesized during a pulse as short as 2 min. Huberman and Riggs 19 have shown that the rate of DNA replication in Chinese hamster cells can be as much as 2.5 µm min –1 , although most replication seems to occur at rates between 0.5 and 1.2 μm min –1 . Thus as much as 5 µm of DNA could be synthesized during a 2 min pulse. If this DNA were synthesized at the nuclear membrane and then stretched out, it could reach the centre of a nucleus of diameter 5-8 µm and give rise to a false impression of the location of sites of DNA synthesis. 中文
This effect could be ruled out by shorter pulses. We calculated that a pulse of 0.5 min would provide sufficient grains, after an exposure of several months, and the resolution (less than 1.25 μm of DNA synthesized) necessary to distinguish between replication solely at the nuclear membrane and replication elsewhere in the nucleus. 中文
In most experiments, we used Chinese hamster (CHO) cells, which can be easily synchronized. The cells were pulse-labelled for 0.5 min with 3 H-TdR, then washed, fixed, embedded, sectioned and stained. The sections were placed on grids and autoradiographed by standard techniques 20 . After exposure times of several months, the emulsions were developed and the grids were examined by electron microscopy. 中文
Cells from an unsynchronized culture, shown in Fig. 1, illustrate the variety of distribution of grains over the nucleus. The cells in Fig. 1A and B are labelled throughout the nucleus (general pattern), whereas the cells in Fig. 1C and D are labelled predominantly around the nuclear membrane (peripheral pattern). The grains in Fig. 1E are mostly clustered over regions of condensed chromatin, frequently near the membrane. Fig. 1F shows a cell with grains throughout the nucleus but with some concentration around the nuclear membrane. 中文
Fig. 1. Autoradiography of unsynchronized CHO cells exposed to 3 H-TdR for 0.5 min. A and B, grains distributed over entire nucleus. C and D, grains distributed around nuclear membrane. E, clustered grain distribution. F, mixed grain distribution. CHO cells were grown on Petri plates in Joklik-modified MEM (Grand Island Biological Company) supplemented with 7% foetal calf serum and non-essential amino-acids. Pulse-labelling was performed by first adding 5-fluorouridine deoxyriboside (FUDR; Hoffmann-LaRoche) to a final concentration of 1.6 μg ml. –1 to inhibit further biosynthesis of dTTP. After 1 min, 3 H-TdR (51 Ci mmol –1 ; New England Nuclear) was added to 17 μCi ml. –1 . After 30 s the plates were removed from 37℃, and the medium rapidly sucked off. The plates were then washed with ice-cold isotonic saline containing FUDR at 0.1 μg ml. –1 (two changes). Cells were removed from the plates by trypsinization at room temperature in isotonic saline still containing FUDR. The cells were pelleted, then fixed with glutaraldehyde and OsO 4 , dehydrated, embedded in epoxy resin 29 and cut into gold-purple sections. The sections were mounted on grids and autoradiographed by the technique of Caro and Van Tubergen 20 except that the acetic acid stop bath was replaced with distilled water. Exposure time was 3.5 months. The bar in each figure represents 1 μm.
Many grains are found in some cells more than 1.25 μm from the nuclear membrane (central grains), suggesting that replication is taking place at sites away from the membrane. To be certain of that conclusion, however, other possible explanations must be ruled out. The central grains cannot be due to background because few or no grains are found over the cytoplasm. They must be due to incorporation into DNA since no grains are found over G1 or G2 cells (see below), and the grains of isolated nuclei can be removed by DNAase. The central grains could also originate from DNA replication occurring on invaginations of the nuclear membrane either just above or just below the section plane. While this might explain some central grains, it certainly cannot explain most of them. Because the cells are sectioned with random orientations, we can get some idea of the frequency of invaginations which might bring nuclear membrane within 1.25 μm of the section plane from the frequency of invaginations in the nuclear periphery. Although some peripheral invaginations are evident in Fig. 1, they are not so frequent that most of the central area could be 1.25 μm or less from an invagination. 中文
In some preliminary experiments we used HeLa cells, which have much more regular nuclei: central grains were also found over their nuclei (Fig. 2). Even serial sections up to 0.5 μm apart through a single cell all showed central grains. The change in pattern from general labelling in early S to peripheral labelling in late S suggests that general labelling is not an artefact of sectioning. Finally, the central grains cannot be accounted for by internal membranes within the nucleus. Such membranes have never been seen by other electron microscopists; apart from invaginations of the peripheral membrane we could see none in our sections. 中文
Fig. 2. Autoradiography of a HeLa cell exposed to 3 H-TdR for 1.5 min. The cells were grown in suspension culture in Joklik-modified MEM (Grand Island Biological Company) supplemented with 5% horse serum. Pulse-labelling was done by adding 3 H-TdR (20 Ci mmol –1 , New England Nuclear) to a concentration of 20 μCi ml. –1 . Exactly 1.5 min after addition of 3 H-TdR, the pulse was terminated by addition of an equal volume of ice-cold isotonic saline containing 1% glutaraldehyde. Subsequent processing was as in Fig. 1. Exposure time was 9.5 months. The bar represents 1 μm.
To express quantitatively the proportion of cells showing peripheral labelling, we have modified the method of Williams and Ockey 16 . We defined central grains as those further than 1.25 μm from the nearest nuclear membrane, and peripheral grains to be those closer than 1.25 μm to the nuclear membrane. The “central activity” of a cell section was calculated as the ratio of the fraction of central grains: the fraction of nuclear area which was central. Nucleolar areas and grains were excluded because the nucleolus has less DNA than the rest of the nucleus. A histogram showing the frequency of various central activities for an unsynchronized cell population is shown in Fig. 3 (a central activity of 1.0 implies an equal concentration of grains over central and peripheral areas). 中文
Fig. 3. Distribution of grains over unsynchronized CHO cells exposed to 3 H-TdR for 0.5 min. Prints at about×15,000 magnification were made of autoradiograms of individual randomly chosen cells similar to and including those in Fig. 1. A line was drawn within each nucleus which was at all points 1.25 μm away from the nearest nuclear membrane. This line divided the nucleus into a central area and a peripheral area. Grains were counted over each area, and the size of each area was measured. In these determinations, grains lying over the nucleolus, and nucleolar areas, were ignored because the nucleolus has so much less DNA than the rest of the nucleus. Grains lying outside the nucleus were also ignored. “Central activity” was calculated as the ratio of the fraction of grains which were central to the fraction of area which was central. Thirty nuclei were measured for the histogram.
Only 13% of the cells in Fig. 3 have central activities of less than 0.25, showing that DNA synthesis can go on at sites away from the nuclear membrane. The nuclear membrane is therefore not essential for DNA replication in Chinese hamster cells. But the fact that many cells do show a predominantly peripheral pattern remains to be explained. 中文
Since previous autoradiographic studies 14-17 had suggested that peripheral labelling occurs only at certain times during the S phase, we decided to try short pulse-labelling of cells synchronized to various times during the cell cycle. We chose “Colcemid” reversal 21 to synchronize the cells. Growing cells were treated with “Colcemid” for a few hours and the cells blocked in mitosis collected by selective trypsinization. These cells (90-100% mitotic) were allowed to continue growing in the absence of “Colcemid”. Within 2 h, 98.5% of the cells completed division, and by 18 h after release from mitosis, more than 70% of the cells had divided again. We preferred this synchronization method to methods involving starvation of nucleotides 21 , because cells recover excellently and normal DNA replication is not affected. 中文
Table 1 and Fig. 4 show the distribution of grains within the nucleus as a function of time after release from mitosis. Peripheral labelling cannot be detected in the early stages of S but becomes predominant in later S. 中文
Table 1. Extent of DNA Synthesis in Synchronized CHO Cells
* Cells were synchronized as in Fig. 4. At 2 h intervals after release from mitosis, 3 H-TdR (20 Ci mmol –1 ; New England Nuclear) was added to each plate (17 μCi ml. –1 final concentration). After 10 min the cells were washed twice with cold isotonic saline, collected by trypsinization at room temperature, allowed to swell in hypotonic medium (2 mM MgCl 2 , 1 mM EDTA, 10 mM KPO 4 , p H 7.7) for 10 min, pelleted, fixed with methanol-acetic acid (3∶1), spread onto subbed glass slides and allowed to dry. The slides were coated with autoradiographic stripping film (“Kodak AR-10”) and exposed for 7 days. After development, the slides were stained with Giemsa stain, then mounted under “Permount”. At least 312 cells were scanned to determine the percent of cells labelled at each time point.
† The total number of grains over each nuclear section was determined for the autoradiographs used in Fig. 4. These figures were divided by the exposure time in months. Extra data for 2 h and 8 h after release from mitosis are included here.
Fig. 4. Distributions of grains over synchronized CHO cells exposed to 3 H-TdR for 0.5 min. A, Early S phase (4-6 h after release from mitosis). B, Late S phase (10-12 h after release from mitosis). CHO cells were synchronized by the “Colcemid” reversal method of Stubblefield. Pulse-labelling (performed at 2 h intervals after release from mitosis), preparation for electron microscopy, and autoradiography were performed as in Fig. 1. Exposure time was 2-4 months. Histograms were prepared as in Fig. 3. For A, eight nuclei pulse-labelled at 4 h and eleven nuclei labelled at 6 h after mitotic release were used. For B, thirteen nuclei labelled at 10 h and sixteen nuclei labelled at 12 h were used.
If DNA is synthesized at the nuclear membrane (or other localized sites within the nucleus), then the DNA should move toward the membrane for replication and away from the membrane afterwards. The pattern of grains after a long pulse or pulse-chase should then be different from that after a simple short pulse. If, on the other hand, DNA replication can take place anywhere in the nucleus, then DNA movement is unnecessary and the grain pattern after pulse-chase labelling could be identical to that after simple pulse-labelling. 中文
To test these possibilities, we first pulse-labelled synchronized cells for 0.5 min during the first half of the S phase (6 h after release from mitosis), and then chased with non-radioactive thymidine to late S phase (12 h after release from mitosis). Although the cells were examined in late S phase, they showed no higher proportion of peripheral labelling than expected for cells 6 h after release from mitosis (compare Figs. 4 and 5). In addition, unsynchronized cells pulse-labelled for 0.5 min show the same proportion of peripheral and general labelling as unsynchronized cells pulse-labelled for 10 min, or pulse-labelled for 0.5 min and then chased for 6.75 h(Fig. 5). In both of the chase experiments, the cells were more labelled after the chase than before, probably because incorporation of 3 H-TdR continues from the cells’ internal pools. 中文
Fig. 5. Distribution of grains over CHO cells after pulse-chase or long pulse-labelling with 3 H-TdR. A, Synchronized cells (see Fig. 4) were pulse-labelled 6 h after release from mitosis as in Fig. 1. Then 30 s after addition of label, the medium was sucked off and the plates were washed, then replaced with medium containing 2.5 μg ml. –1 of TdR. After 6 h the cells were collected by trypsinization and prepared for electron microscope autoradiography as in Fig. 1. Exposure time was 1.5-4 months. B, Unsynchronized cells were pulse-labelled as in Fig. 1 and chased for 6.75 h as in A. Exposure time was 3.25 months. C, Unsynchronized cells were pulse-labelled as in Fig. 1 except that the 3 H-TdR was left in contact with the cells for 10 min instead of 30 s. Exposure time was 1.5-6 months. In all cases, histograms were prepared as in Fig. 3. For A, twenty-four cells were used, whereas for B, twenty-nine cells were used, and for C, thirty-two cells were used.
Thus DNA has a stable location in the nucleus. Because some of the unsynchronized cells must have divided during the chase, our results suggest that DNA bound closely to the membrane in one generation is also bound to the membrane in the succeeding generation. 中文
Because we cannot detect any difference in labelling pattern between a 0.5 min pulse and a 10 min pulse, our results can be compared directly with those of previous investigators 15-17 who have used longer pulse times. Our results agree with most of the earlier results; a general-label pattern is obtained in early S phase and a peripheral one in late S phase 15-17 , and chase experiments show that the nuclear position of DNA, once replicated, is stable 15,16 . 中文
The increased labelling during chase experiments, together with an apparently lower rate of synthesis in early S phase (Table 1), suggests an explanation for the autoradiographic results of O’Brien et al . 13 . In their pulse experiment, generally-labelled cells in early S phase may have been overlooked because they have few grains per nucleus (and a more disperse distribution at that); the increased labelling provided by the chase would have made the generally labelled cells sufficiently obvious to count. 中文
Our results also partially disagree with those of Comings and Kakefuda 14 . They reported that cells synchronized to the beginning of S phase showed solely peripheral labelling. Williams and Ockey 16,22 suggested that Comings and Kakefuda’s results might be due to cell damage during the lengthy nucleotide starvation (24 h in the presence of excess thymidine and then 14 h in the presence of amethopterin) used to synchronize their cells. Indeed, the cytoplasms of the synchronized cells in Comings and Kakefuda’s experiments contain many large vacuoles (indicating serious cell damage), whereas their unsynchronized cells have normal cytoplasm. Why cell damage should result in a peripheral labelling pattern is not clear. The unexpected pattern may be related to the finding 12 that polyoma virus infection causes mouse satellite DNA, which is normally late replicating 23-25 , to replicate before bulk DNA. 中文
Because Comings and Kakefuda’s 14 results must be discounted owing to cell damage, it seems that DNA synthesis can occur throughout the nucleus, and the nuclear membrane is not involved in initiation of replication. We have found, however, that in very early S phase (2 and 4 h after release from mitosis), the rate of replication per cell (measured as the average number of grains per nuclear section over labelled nuclei after a fixed exposure time) is much less than in later S phase (Table 1). This suggests that at the true beginning of S phase, the rate might be so low that we cannot detect labelling, let alone determine whether the pattern is peripheral or general. 中文
Heterochromatin has been shown to be replicated later than euchromatin and is condensed during interphase 26,27 . Our sections and those of others 14-17 , show a preponderance of condensed chromatin attached to the inner nuclear membrane. Thus our results are consistent with the notion 16 that euchromatin is replicated early in S phase and is distributed throughout the nucleus, whereas heterochromatin is replicated in late S phase and is condensed along the inner nuclear membrane and in other discrete areas. 中文
Whereas many of the cell fractionation experiments on the intranuclear location of DNA synthesis 8-11,13 have been interpreted in favour of attachment of replication sites to the nuclear membrane, this is not the only possible interpretation. Equally valid is the possibility that, after cell lysis, special structural properties of the growing points (extensive single strandedness, for instance) might result in their binding more protein, membrane, or other material than bulk DNA. Alternatively, the growing points may actually be attached inside the cell to some material that causes their separation from bulk DNA during lysate fractionation. 中文
As experiments with mammalian 8-11,13 and bacterial 2,3,5 cells used similar techniques and gave the same results, it is possible that the growing point in bacteria may also not be attached to the membrane. 中文
We thank Elaine and Robert Lenk for advice and assistance. This research was supported by grants from the National Science Foundation and the US National Institutes of Health. The electron microscopy was carried out in the Electron Microscope Facility of the Biology Department at MIT. 中文
Note added in proof . Fakan et al . 28 have recently obtained results similar to ours in both autoradiographic and cell fractionation experiments with mouse cells. 中文
( 241 , 32-36; 1973)
Joel A. Huberman, Alice Tsai and Robert A. Deich
Departments of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
Received May 4; revised December 15, 1972.
References: