API, API 682

Should vs Shall vs Must

11 Comments

When I joined the API 682 1st Edition Taskforce in 1991, one of the fine points of writing that I had to learn was to use “shall” instead of “should” when writing standards.  “Shall” just was not, and still is not, a word in my day-to-day vocabulary.  My tendency was to think, say and write “(something or other) should be designed (made of, tested – whatever)” instead of “… shall be …”.

Using “shall” instead of “should” was stressed to the point that I came to believe that API did not allow the use of “should” at all.  Actually, API does allow the use of “should” and explains its use in its “API Document Format and Style Manual”.  Correct usage is also explained in the Foreword to API 682 4th Edition:

“As used in a standard, “shall” denotes a minimum requirement in order to conform to the specification. “

whereas

“… “should” denotes a recommendation or that which is advised but not required in order

to conform to the specification.”

API 682 4th Edition uses the word “should” a total of 183 times!  “Shall” is used 822 times.  When an API standard says you shall do something, it means that you are required to do so.  Seems clear, doesn’t it?

As it turns out, “shall” is not a word of obligation.  The Supreme Court of the United States ruled that “shall” really means “may” – quite a surprise to attorneys who were taught in law school that “shall” means “must”.   In fact, “must” is the only word that imposes a legal obligation that something is mandatory. Also, “must not” are the only words that say something is prohibited. 

Here are some references that say to use the word “must” instead of “shall”:

Frankly, the above references are so long and complex that they were of little help to me but perhaps someone with legal experience can decipher them.  Interestingly, many of the references themselves use “shall” a lot.

It seems that many federal documents are being revised by replacing “shall” with “must” to indicate a requirement.  I wonder if API will soon be doing the same?

11 thoughts on “Should vs Shall vs Must”

  1. Can you provide the case where the Supreme Court ruled that “shall” really means “may” and “must” is the only word that imposes a legal obligation?

      1. That has more to do with how “shall” was used in this case ” As the statutory language is reasonably susceptible to divergent interpretations”. It does not appear to conclude they would rule the same way if their is a clear interpretation.

    2. Castlerock . Police were not liable when the ex took the three daughters and murdered them because the protective order said “shall” and the cops didn’t come when the mom called about him violating the restraining order.

  2. Mr Gordon

    Please

    I have a doubt about the cyclic test with water according to API 682 fourth, I.4.4 numeral edition where are related the steps b) and c) which involve a drop of temperature from 80C to 20C but it doesn´t specify how much time is required to do it.let me to ask if it doesn´t matter the time needed and the way to reach the lower temperature of 20C ?for example if the time required is 5 hours or more to cool liquid in the stuffing box of the test rig ..
    would it be acceptable?

    Thanks a lot

    Mario Correa

    1. Thanks for the question.
      The concept behind the cyclic testing is that there has been an upset of some sort and that the operators are attempting to restore equilibrium. Therefore, the upsets and corrections should happen quickly. Look at Figure I.6 and note that the 1 minute between points d and e is really intended to be typical. After all, the cyclic phase repeats the “upsets” five times. Therefore, you should attempt to reach the lower temperature in one minute. The easiest way to do this is to have a cold water injection ready to use – open a valve and inject the cold water. Close the valve and return to using hot water. Yes, this can be a thermal shock to the seal – that was the intent.

  3. Mr. Buck;
    When working in oil refining, the company I worked at defined the term Shall as it had to be adhered to. Should meant it still had to be adhered to, but if it was not adhered to, an explanation in writing was required.

    1. That is also the API interpretation and intent of “shall”. “Should” is simply a recommendation. Whether or not the API interpretation and definition would hold up in a court of law is another question.

      1. HI Gordon, it’s a pleasure to read through your website and blog. I wanted to ask/comment on the cyclic phases which was brought up in an earlier discussion with Mario Correa (December 2019) however my question relates when doing hot oil testing. Some facilities which have been set up to do this type of testing cannot drop the hot oil temperture in a quick manner, some take about 4-6 hours in order to cool down enough to meet the test parameters per the specification. This means that the full cyclic phases take days to compete. Others which I am familiar with set up an addition loop of cool test fluid which could be switched over quickly and the 5 cyclic phases could be completed in less than a few hours. Is there any issue with the total length of time to drop the process temperature as far as API 4th edition is concerned?

        1. Sorry to take so long, Rob. Virtually all comments received are spam or junk; I don’t understand why people do this. In fact, I was about to revise the blog so that comments were not allowed until I saw yours. Thanks for a good question.

          First, you must realize that my reply is simply my interpretation and not an “official API response”. I still think the temperature cycle is intended to be done quickly. For one thing, the duration of the cycles was never intended to last for days. However, I realize that a rapid temperature cycle is a problem. I also know that the test details were written with little previous experience on which to base such tests. Also, the primary concern in developing the qualification tests was light hydrocarbon emissions and this concern reveals itself in many details that do not follow the light hydrocarbon test analogy.

          I will bring up this question with the 5th Edition taskforce and get the opinion of others. The procedure, if taken literally, is so difficult that it is likely to have never been done exactly as written. I fully expect the test procedures to be significantly re-written by the 5th Edition taskforce — both for clarity and (perhaps) detail.

          Thanks again for your question.

Comments are closed.