KS3科学实验 变量控制 读数误差 图表评估
KS3 的科学课表面上在讲生物、化学、物理，实际上还有一套贯穿三年的实验语言：怎么定变量、怎么读数、怎么把数据变成表格和图像、怎么判断结果能不能信。这套语言在英国 KS3 国家课程里有个正式名字，叫 working scientifically，考试里它很少单独占一整题，却几乎每道实验题都在用。
大纲把 working scientifically 单列成一节
英国教育部（Department for Education）的 KS3 科学国家课程把 working scientifically 写在开头，分四块：科学态度、实验技能与探究、分析与评估、测量。文件要求这一节不能当独立单元教，必须落在生物、化学、物理的具体内容里，讲细胞就做显微镜观察，讲反应就做温度变化。大纲也没有给出必做实验清单，法定的 required practicals 要到 GCSE 才出现，AQA 的 Combined Science: Trilogy（8464）一共列了 21 个，生物 1 到 7、化学 8 到 13、物理 14 到 21。KS3 阶段学校自己决定做哪些实验，考的是方法有没有学会，而不是背过某个实验的步骤。
In England, the Key Stage 3 science programme of study from the Department for Education opens with Working Scientifically as its own section, split into scientific attitudes, experimental skills and investigations, analysis and evaluation, and measurement. The document insists these skills are taught through biology, chemistry and physics content rather than as a separate unit. It sets no fixed list of experiments either; statutory required practicals start at GCSE, where AQA Combined Science: Trilogy (8464) numbers 21 of them across the three sciences.
一句好奇要改成能测的问题
学生常问的问题是哪件外套最保暖。这句话没法直接做实验，因为最保暖没有定义。改成能测的版本：取三个相同的杯子，装同体积同温度的热水，分别裸露、包棉布、包铝箔，每两分钟记一次温度，共十分钟。这时自变量是包裹材料，因变量是十分钟内的温度下降值，控制变量是水量、起始水温、杯子形状、室温、摆放位置。常见的错误是换材料的同时又换了水量，结果出来没法判断是谁造成的差别；另一种错误是把结论提前写进假设，写成包棉布保温因为棉布厚，那是把答案抄进题目，不是预测。
写预测时把条件和结果绑在一起：如果棉布里的空气减缓热量传递，那么包棉布的杯子降温应该最慢。这句话既给了理由，也给出了可以被打败的说法，实验才有意义。做完之后如果数据不支持，不必改口，把不一致的地方写出来，比硬说符合预测诚实得多，也更容易在评估题里拿到分。
A question such as which jacket keeps you warmest cannot be tested as it stands, because warmest is undefined. Turn it into a measurable one: three identical cups, the same volume of water at the same starting temperature, one bare, one wrapped in cotton, one in foil, with the temperature read every two minutes for ten minutes. The material is the independent variable, the temperature drop over ten minutes is the dependent variable, and volume, starting temperature, cup shape, room temperature and position all have to be held constant. The usual mistake is changing the water volume while changing the material, which leaves the result impossible to attribute.
读数能写到几位，由仪器分辨率决定
100 cm3 量筒每一小格是 1 cm3，读到整数就够了；视线要和凹液面最低处齐平，俯视会让读数偏大。温度计每一小格常是 1 摄氏度，写成 23.47 摄氏度等于编造精度。秒表的限制不在刻度而在人，手按停表通常有 0.2 秒左右的延迟，所以反应时间实验里差 0.05 秒的结果说明不了什么。电子天平读到 0.01 g，刻度尺最小分度 1 mm，长度记到 0.1 cm 比较合适。单位写在表头，同一物理量的小数位要统一，原始数据不要涂改，划掉重写并注明原因比擦净重抄可信。
A 100 cm3 measuring cylinder with 1 cm3 divisions justifies a reading to the nearest whole number, taken with your eye level with the bottom of the meniscus; looking down makes the value too large. A thermometer marked in whole degrees cannot honestly report 23.47 degrees. The real limit on a stopwatch is the person holding it, roughly a 0.2 second delay, so a 0.05 second difference in a reaction time experiment means nothing. Units belong in the column heading, decimal places stay consistent down a column, and raw data should be crossed out and annotated rather than rewritten cleanly.
准确、精确、可重复、可复现
大纲把这四个词放在一起写：accuracy, precision, repeatability, reproducibility，题目里也常要求区分。准确是测得的值离真值近，精确是多次测量彼此接近，可重复是同一个学生用同样的方法再做一次得到相近结果，可复现是换一个学生、换一套同型号器材也能得到相近结果。天平没归零，每次称量都多出 0.5 g，这是系统误差，数据整整齐齐却整体偏移；读数时眼睛位置每次都不同，这是随机误差，数据上下散开。
The specification lists accuracy, precision, repeatability and reproducibility in one breath, and exam questions often ask you to tell them apart. Accuracy is closeness to the true value, precision is how tightly repeated readings cluster, repeatability means the same student repeating the method gets similar results, and reproducibility means a different student with a different set of the same equipment does too. A balance that was never zeroed adds 0.5 g to every reading: systematic error, tidy data sitting at the wrong place. Eyes in a different position each time give random error, and the points scatter instead.
结果表要让别人看懂你做了什么
一张合格的结果表至少有自变量一列、因变量一列、重复读数几列和平均值一列，表头写成温度与单位并列的形式，格子里只填数字。平均值保留的位数与原始读数一致，不要因为除了三次就多写两位。明显偏离的读数先别删，查有没有记录上的原因，比如温度计碰到杯壁、砝码掉了一个，能解释就注明，不能解释就留在表里，交给评估部分讨论。删掉不合意的数据，等于让结论替自己挑证据。
平均值不是把三个数一加一除就完事。如果两次读数在 21.0 摄氏度附近、一次是 26.0 摄氏度，先想想第三次是不是碰到了热源，或者读数时温度计刚离开水面。能找出原因就在表下注明并说明是否采用；找不出原因就保留三个数照算平均，再在评估里写这条数据让平均值的可信度下降了多少。
A workable results table carries one column for the independent variable, one for the dependent variable, several for repeat readings and one for the mean. Headings state the quantity with its unit, and the cells hold numbers only. Means keep the same number of decimal places as the raw readings. Before deleting an odd point, look for a recording reason such as the thermometer touching the cup wall or a dropped mass; note it if you can explain it, keep it in the table if you cannot, and discuss it under evaluation instead.
折线还是柱状，看变量是连续还是分类
自变量是连续量，比如时间、温度、长度、质量，用折线图或散点图；自变量是类别，比如材料种类、溶液名称，用柱状图。坐标轴标清物理量与单位，刻度要均匀，图最好占满坐标纸三分之二以上。画线时让点分布在线的两侧，而不是把每个点用折线硬连起来，硬连只适合随时间连续记录的曲线。用图线读中间值叫内插，比较可靠；往两端延长去猜没测过的值叫外推，误差会被放大，题目里要求说明这一点的局限。
刻度怎么定，看数据范围。温度从 20 摄氏度到 40 摄氏度，纵轴从 0 起步会把变化压扁成一条平线，从 15 或 18 起步更能看出斜率，但要在轴上断开的位置做出标记，不能悄悄跳。坐标轴标签漏掉单位是最常见的扣分点，写了温度的 C 却没写摄氏度，读数的人只能猜。
Plot a line or scatter graph when the independent variable is continuous, such as time, temperature, length or mass, and a bar chart when it is categoric, such as material or solution name. Axes need the quantity and its unit, the scale must be even, and the graph should fill most of the grid. Fit a line that lets points sit on both sides instead of joining dot to dot, which only suits a continuous record over time. Reading between plotted points is interpolation and is fairly safe; extending the line beyond them is extrapolation and magnifies error.
结论要念数字，评估要找误差来源
只写温度越高反应越快拿不到过程分。把数据念出来：水温 20 摄氏度时十分钟下降 6.5 摄氏度，40 摄氏度时下降 18.0 摄氏度，降幅约为前者的 2.8 倍，与预测一致。然后才判断预测有没有被支持。评估要看两头：随机误差靠增加重复次数、用数据记录仪自动采样来压小；系统误差靠校准仪器、检查零点、换方法交叉验证来发现。写下次更仔细一点不算建议，要写成能执行的动作，比如每次读数前用同一支温度计在冰水混合物里校准零点。
Writing that hotter water cools faster earns few marks on its own. Quote the numbers first: from 20 degrees the water lost 6.5 degrees in ten minutes, from 40 degrees it lost 18.0 degrees, about 2.8 times as much, which matches the prediction. Only then judge whether the prediction was supported. Evaluation works from both ends: cutting random error by repeating and by logging automatically, catching systematic error by calibrating, checking the zero and cross-checking with another method. Being more careful next time is not an improvement; calibrating the thermometer in ice water before every reading is.
风险这一段，写控制措施而不是口号
大纲把 evaluate risks 写进科学态度，意思是识别危险、判断风险大小、给出降低风险的做法。用本生灯的危险是明火与烫伤，控制措施是长发扎好、护目镜戴稳、先点安全火焰再调成蓝焰、加热时管口不对着人。用稀盐酸的危险是溅入眼睛，措施是减少用量、戴护目镜、手边备清水。学校层面的具体操作规范由 CLEAPSS 这类机构给出，卷面上要写的是一对危险加措施，一句注意安全不算回答。
The programme of study puts evaluate risks under scientific attitudes, which means naming the hazard, judging how serious the risk is and stating what reduces it. With a Bunsen burner the hazards are the open flame and burns, and the controls are tying long hair back, wearing goggles, lighting on the safety flame before turning it blue and never pointing the barrel at anyone. With dilute hydrochloric acid the hazard is splashing into eyes, and the controls are smaller volumes, goggles and water within reach. CLEAPSS guidance governs the detail in school; the exam wants the hazard and control pair.
两个用日用品就能复现的实验
弹簧伸长：把弹簧竖直挂好，依次挂 50 g、100 g、150 g 砝码，每次量总长度并减去原长得到伸长量。质量是自变量，伸长量是因变量，弹簧本身与悬挂方式是控制变量。别超过弹性限度，挂太重之后弹簧回不去，后面的点全部作废。保温实验用三个相同的杯子，装同体积同温度的热水，分别裸露、包棉布、包铝箔，每两分钟记一次温度，十分钟后画出三条降温曲线比斜率，斜率的绝对值就是散热快慢。
两个实验都有个容易忽略的细节：读数的时间点要卡准。弹簧实验要等砝码静止再量长度，晃动时量的偏长；保温实验的水温要在同一位置读，贴着杯壁和插到中心能差半度。把时间点写进方法里，别人照着做才能得到接近的结果，这就是大纲所说的可复现。
For spring extension, hang the spring vertically and add 50 g, 100 g and 150 g masses in turn, measuring the total length each time and subtracting the original length. Mass is the independent variable, extension the dependent, the spring and the way it hangs are controls. Stay inside the elastic limit; overload it and the later points are worthless because the spring does not return. The insulation test uses three identical cups of water at the same volume and temperature, one bare, one in cotton, one in foil, read every two minutes for ten minutes before comparing the slopes of the three cooling curves.
这些动作在 GCSE 会直接换成分数
AQA 的 Combined Science: Trilogy（8464）在 Practical assessment 一节写明，书面卷里基于这 21 个 required practicals 出的题至少占整门资格总分值的 15%。变量识别、表格规范、图像解读、误差评估这些动作因此在六份卷子里反复出现，而且常被放进没见过的情境。KS3 阶段把表格和图像练成习惯，到了 GCSE 只是换一个陌生的实验背景，动作本身没有变。
AQA's Combined Science: Trilogy (8464) states in its Practical assessment section that questions drawing on the 21 required practicals account for at least 15 per cent of the overall marks. Identifying variables, tabulating results properly, reading graphs and evaluating error therefore reappear across all six papers, often inside an unfamiliar context. A student who built those habits at KS3 meets the same moves at GCSE with only the scenario changed.
失分最多的是这几处
阅卷时问题很集中：变量只写自变量和因变量，控制变量一栏空着；单位塞进每个格子，表头反而没有；结论一句定性描述，一个数字都不引用；改进建议停在多做几次，却没说多做几次能解决哪一类误差。还有一类更隐蔽，平均值被四舍五入成整数，后面比较时把真实差别抹掉了。这些都不是知识漏洞，是书写习惯没建立起来，改起来也快。
Marking scripts show the same shortfalls repeatedly: only the independent and dependent variables are named while controls are left blank, units are scattered through the cells instead of the headings, conclusions stay qualitative with no data quoted, and improvements stop at repeating the experiment without saying which kind of error that would fix. A subtler one is rounding a mean to a whole number and erasing the difference the comparison depended on. None of these is a gap in knowledge; they are habits, and habits change quickly.
在家能练的三件小事
用厨房秤称同一袋糖五次，记下读数，算平均值和极差，体会什么叫精确但不一定准确。用手机秒表测十次接尺子的反应时间，看自己的数据散在什么范围，理解为什么单次数据不能当结论。用量杯量同一杯水，分别从正面、俯视、仰视读一次，看偏差有多大。每次做完写一行结果表、一句带数字的结论、一条能执行的改进，这个格式与 KS3 卷面要求一致。
Weigh the same bag of sugar five times on a kitchen scale and calculate the mean and the range, which shows how precise readings can still be inaccurate. Time ten attempts at catching a falling ruler with a phone stopwatch and see how widely your own results scatter, which explains why one reading proves nothing. Read the same beaker of water from the front, from above and from below, and compare the three values. Finish each trial with one table row, one conclusion containing a number and one improvement you could actually carry out.
DK汽车大百科 升级修订版 儿童科普百科¥372.1📱 去京东购买京东精选
需要一对一辅导？Need One-on-One Tutoring?
A-Level / GCSE / IB / KS2-KS3 全科辅导
16621398022 同微信
Visit aleveler.com
或截图后用「扫一扫 → 相册」选择本图