Unable to duplicate problem: Using `[t]` option on tabular affect paragraph spacing

Question

I am having trouble duplicating a problem that is mentioned in the comments in Definitions for equations where it is mentioned that with the [t] option on tabular

spacing becomes bad at the end

I did not originally have the [t] option in the solution, but a comment alterted me to the fact that one gets better spacing just before the tabular with the [t] option and that part seems to be confirmed by the example in the MWE (first tabular uses [t] and second does not).

Questions:

• When is it recommended to use and not use the [t] option in tabular?
• Under what conditions is the spacing affected by the use of the [t] option and when is there no affect of paragraph spacing?

Code:

\documentclass{article}
\usepackage{amsmath}% http://ctan.org/pkg/amsmath
\usepackage{array}% http://ctan.org/pkg/array
\usepackage{lipsum}
\begin{document}
[
    dN/d(d_{p}) = kd_{p}^{\delta},
]
where:
\begin{tabular}[t]{r@{}l}
 $dN$     &: number of particles per unit water volume in the size range $d p$  to  $[d p +d(d p )]$ \

 $k$      &: (= constant) depending on the particle concentration \
 $\delta$ &: $(< 0)$  descriptor of the distributions' spectral slope. 
\end{tabular}
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
[
    dN/d(d_{p}) = kd_{p}^{\delta},
]
where:
\begin{tabular}{r@{}l}
 $dN$     &: number of particles per unit water volume in the size range $d p$  to  $[d p +d(d p )]$ \

 $k$      &: (= constant) depending on the particle concentration \
 $\delta$ &: $(< 0)$  descriptor of the distributions' spectral slope. 
\end{tabular}
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
\end{document}

Answer 1

The spacing is clearly bad: the baseline distance between the bottom of the tabular and the next paragraph is different from the other baseline skips.

The problem becomes very evident if you set differently \lineskip (the glue that's inserted between to lines who happen to be “too near” to each other.

\documentclass{article}

\usepackage{amsmath}% http://ctan.org/pkg/amsmath
\usepackage{array}% http://ctan.org/pkg/array
\usepackage{lipsum}

\begin{document}
\[
    dN/d(d_{p}) = kd_{p}^{\delta},
\]
where:

\begin{tabular}[t]{r@{}l}
 $dN$     &: number of particles per unit water volume in the size range $d p$  to  $[d p +d(d p )]$ \\                                       
 $k$      &: (= constant) depending on the particle concentration \\
 $\delta$ &: $(< 0)$  descriptor of the distributions' spectral slope. 
\end{tabular}

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

\lineskip=2cm

\[
    dN/d(d_{p}) = kd_{p}^{\delta},
\]
where:

\begin{tabular}{r@{}l}
 $dN$     &: number of particles per unit water volume in the size range $d p$  to  $[d p +d(d p )]$ \\                                       
 $k$      &: (= constant) depending on the particle concentration \\
 $\delta$ &: $(< 0)$  descriptor of the distributions' spectral slope. 
\end{tabular}

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

\end{document}

Answer 2

Every box has a top, bottom, and baseline. LaTeX places the baselines \baselineskip apart, or adds \lineskip between a consecutive bottom and top if the glue inserted to preserve baselineskip spacing would be less than \lineskiplimit (in other words, whichever is larger).

The following shows the top, bottom and baseline for your example tabulars. The third example uses \struts to compensate for the lack of decenders and ascenders in the text.

\documentclass{article}
\usepackage{amsmath}% http://ctan.org/pkg/amsmath
\usepackage{array}% http://ctan.org/pkg/array
\usepackage{lipsum}

\fboxsep=0pt

\begin{document}
\[
    dN/d(d_{p}) = kd_{p}^{\delta},
\]
where:

\noindent\fbox{\rlap{\begin{tabular}[t]{r@{}l}
 $dN$     &: number of particles per unit water volume in the size range $d p$  to  $[d p +d(d p )]$ \\                                       
 $k$      &: (= constant) depending on the particle concentration \\
 $\delta$ &: $(< 0)$  descriptor of the distributions' spectral slope. 
\end{tabular}}\rule{\textwidth}{.5pt}}

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

\[
    dN/d(d_{p}) = kd_{p}^{\delta},
\]
where:

\noindent\fbox{\rlap{\begin{tabular}{r@{}l}
 $dN$     &: number of particles per unit water volume in the size range $d p$  to  $[d p +d(d p )]$ \\                                       
 $k$      &: (= constant) depending on the particle concentration \\
 $\delta$ &: $(< 0)$  descriptor of the distributions' spectral slope. 
\end{tabular}}\rule{\textwidth}{.5pt}}

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

\[
    dN/d(d_{p}) = kd_{p}^{\delta},
\]
where:\strut

\noindent\begin{tabular}{r@{}l}
 $dN$     &: number of particles per unit water volume in the size range $d p$  to  $[d p +d(d p )]$ \\                                       
 $k$      &: (= constant) depending on the particle concentration \\
 $\delta$ &: $(< 0)$  descriptor of the distributions' spectral slope. 
\end{tabular}

\strut
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx xxxxxx.

\end{document}